Novus Portus & Hæstingaport at Old Winchelsea

Claudius Ptolemy recorded an otherwise unknown Roman port on the south coast of England. He named it ‘καινòς λιμην’, Greek for ‘new port’. It is usually referred to by its Latin equivalent ‘Novus Portus’ (or being Latin, ‘Portus Novus’ is just as valid). A dozen or so locations have been proposed for Novus Portus – listed below – but there is no consensus on which is probably right. In this paper, we will explain why we think Novus Portus was at Old Winchelsea, and that it became Hæstingaport in Anglo-Saxon times.

Ptolemy

The reference to Novus Portus appears in Ptolemy’s ‘Geographike Hyphegesis’, a mid-2nd century gazetteer of over six thousand places known to the Romans. Each place is listed with quasi longitude/latitude coordinates. They are generally referred to as ‘stations’. Romano-British placenames were Latin, so Ptolemy transliterated or translated them to Greek. It is normal practice to revert them back to Latin. The British coast is split into three sections, east, west, and south. The south coast stations, west to east, are:

 

Ocrinum Promontorium	12 00, 51 30
Cenionis fl. ostia	15 00, 51 54
Tamari fl. ostia	15 40, 52 10
Isacae fl. ostia	17.00, 52 20
Alauni fl. ostia	17 40, 52 40
Magnus Portus	19 00, 53 00
Trisantonis fl. ostia	20 20, 53 00
Novus Portus	21 00, 53 30
Cantium Promontorium	22 00, 54 00

 

In principle, these coordinates are like modern longitude/latitude, only with a different meridian. They can be transposed onto a grid to create a map. In doing so (Figure 1), Britain is grossly distorted, most notably insofar as Scotland is rotated 90° clockwise. The rest is not much better. William Flinders-Petrie, the famous Egyptologist, starts his paper on Ptolemy’s Britain by saying the coordinates are: “more or less erroneous and incomprehensible”. So, what has gone wrong?

Ptolemy’s base was six reference points in Britain whose latitudes were calculated from the length of the longest day at that place. They are overstated, especially at and above the Humber. This is the probable cause of floppy headed Scotland. It is presumably because, being in Britain, longest days had to be estimated due to cloud cover. However, the two southern longest day references - the Isle of Wight and London - are relatively accurate. Something else has gone wrong with the south coast coordinates.

There was no way to measure longitude until the 18th century, so the other coordinates were triangulated from the longest day reference points using distances along roads and coasts. The technique is good for the times. It works fairly well for Ptolemy’s coordinates in Continental Europe, but not for Britain.

Gordon Hills pointed out 150 years ago that Ptolemy wrongly assumed that the Roman-known-world was half of the real world, so he divided it into 180° whereas he should have divided it into 130°. Hills tried to correct this error by multiplying Ptolemy’s longitudes by 130/180, thereby reducing them by approximately 28%. Henry Bradley spotted that Ptolemy got the whole of Britain two degrees too far north. The longitudes should therefore be adjusted by a further 2% or so because the distance between Great Circles narrows as latitudes increase. He also thinks that Ptolemy wrongly calibrated his longitudes at 500 stadia per degree at the equator, whereas they should be 600 stadia. He therefore proposed that Ptolemy’s longitudes should be reduced by 20%, instead of Hills’ 28%. We are not convinced he is right. Ptolemy might have been referring to the length of one degree at his home in Alexandria, which is 31° above the equator where one degree would be close to 500 stadia. Regardless, while they are a small improvement, these adjustments only change the scale and aspect ratio of the coordinates. Other errors create the worst distortions.

Ptolemy’s degrees of longitude represent at least 35 miles on average to the east of Southampton but no more than 27 miles on average to the west, a 30% difference. The only variable that could cause this sort of discrepancy is the distances. The Romans would not have had much trouble measuring their famously straight roads, but the road network in Britain was skeletal when Geographike Hyphegesis was being compiled. Most of the distances would have been taken from coastal sailing distances. They must be materially and inconsistently wrong.

In summary, Ptolemy’s raw coordinates are wildly wrong. Various adjustments have been tried. None of them work well enough to confidently identify Novus Portus or any of the other unknown stations. Historians have tried to do so using other techniques. We will look at these next, then return to propose a better technique to correct Ptolemy’s coordinates.

Review of Novus Portus candidates

Experts, bar Flinders-Petrie, have hitherto tried to locate Novus Portus using toponymy or adjustments to Ptolemy’s mapping technique. None of them have been successful enough to prevent successors trying to make improvements. We will briefly run through them.  

As far as we know, William Lambarde was the first to propose a location for Novus Portus, in his 1570 book ‘Perambulation of Kent’. He reports that his acquaintance William Talbot told him about the origin of the placename ‘Shipwey’: “it was called Shipwey bicause it lay in the way to the Haven where the ships were woont to ride. And that haven taketh hee to bee the same, which of Ptolomie is called Novus Portus : of Antoninus, Limanis, of our Chroniclers Limene Mouth, and interpreted by Leland to betoken, the mouth of the river of Rother, which now in our daies openeth into the Sea at Rye, but beforetime at Winchelsey”. So, Talbot is saying that Novus Portus was at the mouth of the Rother, and that it was at modern Winchelsea in Roman times.

William  Camden discusses Novus Portus in his English history reference ‘Britannia’, published 1605. His theory is based on Ptolemy’s name ‘λιμην’, pronounced ‘Limenu’. He says the name is ‘significative’, meaning it is linked with the similar sounding ‘Lemanis’ and ‘Limen’. These, respectively, are Antoninus and Notitia’s name for a port at modern Lympne, and the Old English name for the River Rother. So, he too thinks that Novus Portus was at the mouth of the Rother but reckons that it was at modern Lymne in Roman times. William Holloway uses the same argument in his 1849 book ‘The History of Romney Marsh’.

William Somner’s ‘A treatise of the Roman ports and forts in Kent’, published in 1693, simply states that Novus Portus was at Old Romney, but he offers no explanation. James Elliot, Romney Marsh chief engineer, picked up the argument 150 years later. In an addendum to Charles Roach Smith’s 1852 book ‘Report On Excavations Made On The Site Of The Roman Castrum At Lymne In Kent’, he explains that the Romans constructed a ditch along the route of the Rhee Wall to redirect the Rother to discharge at Old Romney instead of Lympne. It needed a new port which he thinks to have been Novus Portus.

Mark Anthony Lower, the first translator of the Chronicle of Battle Abbey, wrote about Novus Portus in the 1863 edition of the Sussex Archaeological Collections. He argues, like his predecessors, that Novus Portus was at the mouth of the Rother. He starts out agreeing with Camden that it was probably at Lympne but seems to lose the courage of his convictions by concluding: “Whether the Novus Portus of Ptolemy was there, or more westward at Old Romney, is a moot point”.

Laurence Echard, famous for his 1707 English history reference, also wrote ‘The Classical Geographical Dictionary’, published in 1715. It simply states, without any supporting evidence, that Novus Portus was at Rye. John Horsley agrees. In his ‘Britannia Romana’, published in 1732, like many of his predecessors, says that Novus Portus was at the mouth of the Rother, but he thinks it discharged ‘near Rye’ in Roman times. Thomas Reynolds uses this same argument in his 1799 book ‘Iter britanniarum’. None of them provide any supporting evidence.

Nathaniel Salmon’s ‘New Survey of England’, published in 1731, is an attempt to address what he perceives to be errors in Camden’s history. Having explained that Magnus Portus was a harbour at the mouth of the River Test, he then skips Trisanton fl to propose that Novus Portus was at Chichester. He provides no supporting evidence, and this seems the least likely of the candidates. It would mean having three stations within 15 miles, then none for 95 miles.

Charles Bertram’s ‘De Situ Britanniae’ was published in 1748. It falsely claimed to have been written by a 16th century monk, but that does not matter here. He proposed, without any evidence, that Portus Novus was at modern Eastbourne.

Konrad Mannert’s ‘Geographie der Griechen & Romer’, published in 1804, reviews Roman sites on the south coast. He concludes, without explaining his reasoning, that Novus Portus was at Hythe, presumably because it is near Lympne and ‘hythe’ was Old English for ‘small port’.

Rev Edward Turner published a paper about Lewes in the 1849 Sussex Archaeological Collections in which he proposed that Novus Portus was at Newhaven, presumably because, alone among all the candidates, its name is a valid translation of Novus Portus. His also notes that Newhaven was at the end of two Roman roads, which is good evidence that there was a nearby Roman garrison and port/haven. Margary agrees that he is probably right about the Roman roads, although no evidence of them has been found.

Beale Poste’s ‘Britannic researches ; or, New facts and rectifications of ancient British history’, published in 1853, analysed four Novus Portus candidates: Romney, Rye, modern Hastings and modern Pevensey. He rejects Romney on the basis that Lydd was almost surrounded by sea in the 8th century, so Romney would not have been a sheltered haven. He rejects Rye in the belief that it was probably not formed until the coast shifted in the 13th century. He rejects modern Hastings because it never had a port. By a process of elimination, he concludes that Novus Portus was probably at modern Pevensey. In our opinion, his reasons to reject Romney and Rye are faulty, and they are better candidates.

Gordon Hills and Henry Bradley are the earliest historians that tried to locate Novus Portus by correcting Ptolemy’s mapping method. They both made valid points – see above – leading to different conclusions. Hills says that he wanted to avoid being specific about Novus Portus’s location but hints it was at modern Winchelsea on his map. Bradley calculates that each of Ptolemy’s degrees of longitude is 40 imperial miles, to reach the conclusion that Ptolemy’s coordinates for Novus Portus “give the precise longitude of [modern] Hastings”. It would be a reasonable argument if Ptolemy’s degrees of longitude were consistently wrong, but they aren’t.

H F Napper responded to Hills paper in the 1882 Sussex Archaeological Collections Notes & Queries. He proposed that a waypoint is missing from Ptolemy’s list, meaning that longitudes for the others is out by 1° or 2°. He concludes that if Novus Portus is out by 1°, it was at modern Pevensey, if it was out by 2° it was at Rye. 

Thomas Cole, in his 1884 book ‘The antiquities of Hastings’, concludes his argument that Novus Portus was at modern Hastings: “1st, actual examination of the locality shews the former existence of a harbour here, a fact in agreement with both tradition and history. 2nd, the encampment and vestiges of ironworks prove the harbour to have been known to the Romans. 3rd, a Roman harbour called Portus Novus was situated at this very part of the Coast. These three considerations put together seem to lead inevitably to the conclusion that Hastings was a Roman Port, and that Portus Novus was the name of Roman Hastings.” As far as we know, he invented the evidence for his first two points, so his third argument is invalid.

Thomas Glazebrook Rylands picked up where Hills and Bradley left off in his 1893 book ‘The Geography of Ptolemy Elucidated’. He resolved some of Ptolemy’s mapping errors by correcting London’s coordinates and using it as the datum from which the other distances were triangulated. This adjusted the longitude of the eastern south coast stations by about half a degree, and the western stations by about 1°. Using this technique, he remapped Novus Portus to Dungeness.

Relatively few papers about Ptolemy’s south coast stations appeared in the 20th century. Flinders Petrie, writing in 1917 about a technique we will explain below, suggests that Novus Portus was either at Pevensey or Rye. Albert Rivet’s ‘The place-names of Roman Britain’, published 1979, says that Novus Portus ‘almost certainly’ referred to Dover, although he provides no supporting evidence and he is ‘almost certainly’ wrong, for reasons we return to below. Shirley Read, unsurprisingly given the title of her 1992 book, ‘Bygones of Bexhill’, reckons that Novus Portus referred to Bexhill, but offers no supporting evidence.

Coming to more recent works, Andreas Kleineberg updated Ptolemy’s map for his 2012 book ‘Europa in der Geographie des Ptolemaios’. He supports Turner’s theory that that Novus Portus was at Newhaven.  Johan Åhlfeldt places Novus Portus at Portslade as on his immense digital map of Roman Europe. The website roman-britain.co.uk agrees. None of them provide any justification or evidence.

In 2015, Bernard Leeman said that Old Winchelsea: “may have been the Roman port of Portus Novus mentioned by Ptolemy”. We think he is right, for reasons we explain below, but it is difficult to give him the credit for the discovery because he says that Edward Brayley was his source, but Brayley says that Novus Portus was at Rye. Brayley cites Horsley as his source, and Horsley does indeed say Novus Portus was ‘near Rye’.  

The most recent academic analysis, as far as we know, was produced by Corey Abshire, et al, in their 2020 paper ‘Ptolemy’s Britain and Ireland: A New Digital Reconstruction’. They used a new ‘flocking’ technique to work out that Novus Portus was probably at Lympne. So, somewhat dispiritingly, after five hundred years and dozens of investigations, the experts have ended up exactly where they started, and we think they are still wrong.

To summarise, the similarity between Ptolemy’s ‘Limenu’ and ‘Limen’, the Old English name of the Rother, leads most experts to think that Novus Portus was at mouth of the Rother. The similarity between ‘Limenu’ and ‘Portus Lemmanis’, Lympne’s Roman name, leads many of these to further deduce that the Rother discharged at Lympne. If so, they reason that Novus Portus was at modern Lympne. Others think that the Rother discharged at Romney or Rye in Roman times, so place Novus Portus there instead. Dover is the best supported of the other candidates because Novus Portus can mean ‘new port’ and Dover is the only known new port on the south coast at the time Geographike Hyphegesis was being collated. Cases have also been made for ‘Newhaven’, a viable translation of ‘novus portus’, Hythe which means ‘small port’, Portslade which sounds like a port, Eastbourne, Pevensey, Bexhill, modern Hastings, Old Winchelsea, and Dungeness.

Correcting Ptolemy’s coordinates

To recap, the south coast stations, west to east, are: Ocrinum Promontorium, Cenionis fl. ostia, Tamarus fl. ostia, Isacæ fl. ostia, Alaunus fl. ostia, Magnus Portus, Trisantonis fl. ostia, Novus Portus, and Cantium Promontorium. Their locations are uncertain, but we will propose a technique for them to be calculated.

Latin ‘fl. ostia’ means ‘river mouth’, so five of the stations are estuaries. The ‘portus’ part of Magnus Portus and Novus Portus can mean ‘port’, ‘harbour’, ‘bay’, or ‘haven’. Ocrinum Promontorium was a south coast promontory so it would have had a sheltered haven on its eastern, lee, side. We think Cantium Promontorium referred to North Foreland (see below), making it adjacent to the sheltered haven of Pegwell Bay.

So, all Ptolemy’s south coast stations were havens. Some were also harbours or bays or ports, but the only thing they all have in common is that they were havens. Indeed, all Ptolemy’s other British coastal stations were havens too. This implies, as Flinders-Petrie pointed out a hundred years ago, that the coastal stations were sheltered refuges where ships might safely moor. He goes on to say: “The first point in the coasting is that the places along a given piece of coast are at equal distances. Remembering that Ptolemy usually only states the nearest 10, and never less than 3’. any distances agreeing within such limits are to be considered equal. This equal spacing probably results from statements of day's sailing between ports.” In other words, he proposed that the south coast stations were separated by one day’s coast hugging sailing, with the implication that the stations were where ships might safely moor overnight.

As far as we know, the only south coast stations with supporting locational evidence are Tamari, the Roman name of the River Tamar, and perhaps Isacæ which sounds like it derives from ‘iasc’, the Brythonic root of the River Exe and River Axe. Lots of locations have been proposed for the others, but with no consistency. Flinders-Petrie had the only credible strategy for calculating the locations of the stations, but it looks unsound to us, and he missed a station which messed up his calculations anyway.

Even the three stations about which there is some consensus - Tamari fl ostia at Plymouth, Isacae fl ostia at Exmouth, and Alauni fl ostia at Weymouth - are specious. Plymouth to Exmouth is 33 lateral miles. Ptolemy’s coordinates give them a longitudinal difference of 1.33 degrees, making 24.63 miles per Ptolemaic degree. Exmouth to Weymouth is 47 lateral miles. Ptolemy’s coordinates give them a longitudinal difference of 0.66 degrees, making 71.21 miles per Ptolemaic degree. So, they have a discrepancy of nearly 200%. It looks irreconcilable, but we are going to try .

It seems to us that the worst longitudinal discrepancies are where the stations are either side of a big headland. Ptolemy never visited England and there were no maps for him to work from, so he had no idea how the coast looked. He had no alternative than to assume the coastings between stations were straight. If coastings are substituted into the equations for miles per Ptolemaic degree, the results are far more consistent. Most are in the mid-40s. The extremes are 38 and 52, within the bounds of sailing speed measurement errors using the technology of the day.

This correction identifies the stations west of Novus Portus as:

Ocrinum Promontorium	Newlyn/Penzance
Cenionis fl. ostia	Fowey
Tamari fl. ostia	Plymouth
Isacae fl. ostia	Seaton
Alauni fl. ostia	Weymouth
Magnus Portus	Solent
Trisantonis fl. ostia	Newhaven/Seaford bay

 

These stations not only match Ptolemy’s adjusted coordinates but they are consistent with history. Newlyn/Penzance and Fowey were Celtic settlements, presumably fishing harbours, so they were viable havens. Plymouth is at the mouth of the Tamar river which took its name from the Tamari. Seaton is at the mouth of the River Axe which took its name from Brythonic ‘iasc’. Flinders-Petrie reports that there was a “station of the 2nd Legion 10 miles from Alaunus, and the station of Dorchester is 7 miles from Weymouth”, so Alaunus fl ostia probably referred to Weymouth. Latin ‘magnus portus’ can mean ‘great haven’, and the Solent was the biggest haven on the south coast. Margary reckons that an important Roman road ended at Newhaven/Seaford bay, so it was probably a harbour and haven in Romano-British times.

Before extrapolating what this means for Novus Portus, we have to think about Cantium Promontorium. Nearly everyone assumes it refers to South Foreland. But the smooth rounded curve of South Foreland is not a promontory in any normal sense. Cantium Promontorium is more likely to refer to North Foreland, which is very clearly a promontory. Indeed, until the mid-19th century, Cantium Promontorium was always assumed to mean North Foreland. We cannot find the root of the change in received wisdom, but we think it is unfounded. If Cantium Promontorium did refer to North Foreland, Ptolemy’s station would be at Pegwell Bay.

There are 82 coasting miles and 1.66 Ptolemaic degrees between Seaford and Pegwell Bay, giving an average of 49 miles per Ptolemaic degree. It sounds credible, towards the upper end of the 38/52 bounds calculated above. Using these bounds, Novus Portus is somewhere between 25 and 34 coasting miles coasting miles east of Seaford. This places it between modern Hastings and modern Rye, probably closer to the latter because it is closer to the upper bounds and because the coast between modern Hastings and Fairlight had a rocky lee shore to be avoided at all costs when sailing in ships with no centreboard or daggerboard.

If Novus Portus was on the stretch of coast between Hastings and Rye, or indeed between Combe Haven and Romney, it is unknown in Romano-British history. This is what we will investigate next.

The Rochester Roman road

There is a compelling clue about Novus Portus’s location that has been hiding in plain sight: the Rochester Roman road. Roman roads were built to move soldiers between garrisons and/or major population centres. Those that went to the coast serviced coastal garrisons. Roman coastal garrisons often overlooked nearby or integrated ports. This applies to Ermine Street, Pye Road, both ends of Watling Street, Akeman Street, Stane Street, Portway and the Fosse Way, all built in the first decades of the Roman occupation. The Rochester road is the only other major first century Roman road that terminated at the coast. If it is like the others, it terminated at a garrison, and that garrison probably overlooked a port or haven, and that port or haven is in the right vicinity to have been Novus Portus.

Ivan Margary, doyen of Britain’s Roman roads, traced the route of the Rochester road (R on Figure 2) to Sedlescombe. The main street through Sedlescombe is named ‘The Street’, nearly always an indication that it is on the route of a Roman road. Margary’s route has been corroborated by sections of the road found by HAAG and David Staveley as it passed through Footlands and north Sedlescombe.

There is a general assumption that the Rochester road crossed the Brede at Sedlescombe, climbed onto the Hastings Ridge and ended at modern Hastings. It didn’t. No evidence has been found of a Roman road anywhere on the Hastings Ridge or anywhere in the vicinity of modern Hastings. On the contrary, HAAG have discovered that it forked south of the Brede with one branch going to the immense iron workings at Beauport Park, the other going to modern Winchelsea (W on Figure 2).

Roman roads were built to service garrisons. Therefore, there was a Roman garrison at modern Winchelsea. This is corroborated by Wickham Manor just outside modern Winchelsea. ‘Wickham’ is an Old English placename meaning ‘settlement at a vicus’. If there was a vicus at Wickham, there was a Roman garrison at modern Winchelsea.

It is salient to note that the Rochester road is different from all the other early Roman roads. Indeed, it is different from the later ones too. To explain why, it is important to understand the nature of the Roman invasion.

Resistance to the Roman invasion was in the north, east and west. Southern England seems to have welcomed the Romans. It soon converted into peaceful Romanised states known as ‘civitas’, run by Celtic client kings. Indeed, it has been argued that the south of England was effectively a Roman province long before the invasion because they traded extensively with Romans and had adopted many Roman customs. As each of the other tribes was pacified, it too was converted into a civitas.

The civitas capitals, each with a Roman garrison, were at the old Celtic tribal capitals. The Romans formed a new capital at London and joined it to the southern civitas capitals by paved Roman trunk roads. There were three on the south coast: Noviomagus at Chichester (N on Figure 2), capital of the Regni; Durnovaria at Dorchester (D), capital of the Durotriges; and Isca Dumnoniorum at Exeter (E), capital of the Dumnonii.

Three more Roman roads reached the south coast before Geographika Hyphegesis was collated. They went to Winchelsea, Dover and modern Richborough (Roman Rutupiæ). The earliest, Rutupiæ, was joined to Canterbury and London by Watling Street, the first major Roman road in Britain. Canterbury was the civitas capital of the Cantii tribe, so Watling Street had much the same purpose as the roads to Chichester, Dorchester and Exeter. The extension to Rutupiæ at was built because it was the port through which Roman legions and their supplies arrived in Britain in the first century.

Dover, known to the Romans as ‘Dubris’, was developed by the Classis Britannica at the Celtic harbour of ‘Dubras’ in the early second century. There is no obvious rationale. Rutupiæ was just 11 miles away. It had been servicing the cross-Channel traffic since the invasion and the Romans had already connected it by road to Canterbury and London. Rutupiæ continued to be used through the Roman occupation, so the motivation to build Dubris was not that Rutupiæ had become unusable. The Watling Street branch to Dubris was contemporary with the port development, so it was not the motivation to develop the port either. Nor was it increased traffic: second century traffic was less than the first, and it would have been quicker and easier to install freight handling facilities at Rutupiæ. We suspect that Dubris was built because the Classis Britannica needed a deeper harbour and/or a south facing harbour entrance to handle large cargo vessels.

Whatever the reason for its construction, Dubris was a new port at the time when Ptolemy was writing Geographike Hyphegesis, so it attracts lots of supporters as the site of Novus Portus. It is spurious, we think:

• Most of Ptolemy’s information about Britain seems to have been collated in the first century. Dubris was not connected to the road network at the time, so it would have been a relatively minor harbour or haven. And old, as evidenced by the discovery there of the oldest largely intact ship in the world, dating to 1500 B.C. In other words, it was not a ‘new port’, ‘new haven’, or ‘new anything else’ when Geographike Hyphegesis was being collated, so it is unlikely to have been Novus Portus.
• The Roman name Dubris is a Latinised version of Dover’s Celtic name Dubras, so it was always known as one or the other. If it had a recognised name, Ptolemy had no reason to refer to it as Novus Portus.
• Dubris was only a couple of miles from Cantium Promontorium, whereas Ptolemy’s coordinates for Novus Portus locate it a third of the way to Southampton. His longitudes might lack precision, but none of the others are that wrong.
• Dubris was close to Cantium Promontorium and far from Southampton, making it inconsistent with Flinders-Petrie’s theory that the south coast stations are separated by roughly a day’s sailing.

So, all the pre-Geographike Hyphegesis Roman roads in Britain, bar the Rochester road, were built to connect garrisons at civitas capitals, conflict centres, or major population centres. This did not apply to modern Winchelsea which was in the peaceful south and 40 miles from the nearest civitas capital or population centre. The Rochester road and the garrison it served, must have had some other purpose.

In summary, modern Winchelsea is on the right stretch of coast to have been Novus Portus. It had a garrison. Most coastal Roman garrisons overlooked ports or havens, but there is no historical record of one near modern Winchelsea. If we can work out the purpose of the garrison, it might give a clue about whether it had a nearby port or haven, and what happened to it. That is what we will look at next.

Novus Portus at Old Winchelsea

‘Novus’ means ‘new’, ‘Portus’ means ‘harbour’, ‘bay’, ‘haven’ or ‘port’. The first three of these are geographic features that develop over geological time. They could not be new in the 2nd century. But it seems unlikely that ‘Novus Portus’ was a port either. Nearly all Roman ports serviced the army or major populations centres. This certainly applies to the only well-known first century Romano-British ports, at London, York and Rutupiæ. A port in East Sussex would do neither. True, some Saxon Shore Forts with integral ports were built in the vicinity in the third century, but that was in response to the threat from Germanic raiders which was absent in the first and second centuries.

There are two plausible reasons for a ‘new haven’ or ‘new port’ in Romano-British East Sussex. One is that Novus Portus was ‘new’ in the sense of being ‘newly used’. So, perhaps, a nearby estuary haven got blocked or silted up, so they had to start using a new one. Certainly, the estuaries along the stretch of coast between Fairlight and Lympne were prone to getting blocked by shifting shingle. But a newly used estuarine haven would surely take the name of its river, like thousands of Ptolemy’s other estuarine stations. The alternative is the third reason that Romans built new ports, to facilitate the export of natural resources. We think Novus Portus was this latter.

The Weald had the richest reserves of accessible iron ore in the Roman empire. Enormous quantities of Wealden iron blooms were exported to France to be turned into weapons and armour. They needed to be transferred from estuary barges to sea going ships. Celts had no freight handling ports, so the Romans needed to build at least one new port.

Figure 3 shows the distribution of significant Roman bloomeries in East Sussex based on the Wealden Iron Research Group’s bloomery database. Numerically, most of them are in the High Weald, but volumetrically the three giant Grade 4 sites – Footlands, Oaklands Park and, above all, Beauport Park – are in the Brede basin. According to Henry Cleere, more than 80% of the Weald’s iron ore reserves by volume were in the Brede basin. The Rother basin, Cuckmere basin and Combe Haven basin had roughly 5% each.

Even the smallest of the Wealden iron ore basins, Combe Haven, had a new Romano-British freight handling mini-port, evidenced by the recent discovery of Roman piers at Redgeland. The Cuckmere and Rother basins probably also had new mini-ports of a similar size. The port at the mouth of the Brede, judging by the volume of ore extracted, would have been at least sixteen times bigger.

‘at least’ because the Brede estuary also had the largest number of saltpans on the south coast, not far short of the rest combined. Well, it did in the 11th century according to Domesday, but they were probably of Roman origin. Anglo-Saxons, after all, are not known to have developed any industries in scale and the saltpans were already well developed in 795 when they were gifted to the Frankish Abbey of St Denys (Barker translation with the original placenames): “And by this deed of gift I also grant for the use of the monks there serving God, ports/harbours belonging to me, called Hastingas and Pævenisel, situated on the sea-coast in the same neighbourhood, with saltpans and all things pertaining thereto.” 795 might be 300 years after the initial Anglo-Saxon invasion, but the Hæstingas tribe that occupied the land for most of that time could not have been producing a significant amount of salt or iron, or they would have been overrun long before Offa did so in 771.

The Brede’s salt would have been used for salting fish, any surplus being exported. Salting fish was a labour-intensive business. It would come as no surprise if the port that exported the Brede basin’s iron blooms and salted fish was 20 times the size of the others.

There is a complication. Figure 3’s bloomeries are superimposed on Cleere and Hodgkinson’s interpretation of the Romano-British Romney Marsh coast. It shows the Brede and Rother flowing into an inland lagoon – others show it as a saltmarsh - that discharged into the Channel through a narrow gap at modern Lympne. Other geographic regressions show the Brede flowing into the Rother and the Rother discharging into the Channel at Lympne. If either is right, the Brede’s natural resources would have been transferred from barge to ship at Lympne. If this were so, as many experts propose , albeit for different reasons, Novus Portus could only have been at Lympne.

But extensive excavations at Lympne have not uncovered any archaeological evidence of a first or early second century port. Cunliffe speculates that it might have been destroyed when the third century port was built over it. This seems unlikely. The port that has been excavated at Lympne is too small to have processed the amount of iron and salt coming out of the Brede estuary, and it seems implausible that the Romans would destroy a large port to make a small port. 

James Elliot had a different theory. He proposed that the Rother’s port at Lympne’s was getting blocked by shingle, so he thought that the Romans redirected the Rother south and divided it by building a canal between Appledore and Romney along the route of the Rhee Wall (Figure 4). This led him to propose that the Brede’s iron blooms were shipped upstream to Appledore then along the canal to be exported from Novus Portus at Old Romney. There is a general agreement now that the Rhee Wall was dug no earlier than 13th century. Even if it is earlier, the port at Lympne was still being used in the 3rd century, so the Rhee Wall canal was not dug before Geographike Hyphegesis was written. Either way, Novus Portus was not at Old Romney.

The only plausible alternative is that the Brede discharged into the Channel through a southerly breach in the Camber shingle bar. The Brede’s saltpans provide a clue that this was so. If the Brede discharged at Lympne, its seawater ingress would have been at Lympne. The tide would be weakened as it passed 20 miles of marshland, and the salt would get increasingly diluted by freshwater from the Rother, Tillingham and Brede. One would expect the number of saltpans to reduce towards the south, but the evidence shows the opposite. Indeed, there are no known saltpans on Guildeford Marsh, Walland Marsh or the Tillingham Estuary, through which Lympne’s seawater would have had to pass to get to the Brede. The only other saltpans in the region are at Lydd, Bilsington and Eastbridge, relatively small and around the north and east periphery of Romney Marshes. Therefore, there must have been a southerly gap in the Camber shingle bar to allow seawater ingress directly into the Brede. That gap would be the de facto mouth of the Brede, and the port would have been nearby.

John Peterson has another reason to think there was a southern breach, explained in his 2002 paper ‘Le réseau centurié Kent A’. His main evidence is an Ordnance Survey diagram (Figure 5) based on Green’s 1968 Soil Survey. Peterson says that the brown areas are “decalcified because they have been dry land at most times since the Roman period”, the yellow areas are “more recent sediments deposited in the middle ages”, the pink areas are soils laid on sand/shingle of uncertain age. So, Peterson’s diagram shows a decalcified watershed between the Lympne inlet to the north and an inlet to the south, and it has been there since Roman times. The junction of the brown, yellow and grey shading to the NNE of Rye is near Appledore. This suggests that the Rother also discharged into the Channel through a southerly breach in the Camber shingle bar.

There is another reason to believe that there was a Romano-British port at the mouth of the Brede. The 795 St Denys charter mentioned above specifically says that a portus at Hastingas is included in the gift. There would have been no reason to mention it if it were just a natural harbour, so it must have levied taxes for the transfer of freight. The transfer of freight would have needed manmade piers, quays, warehouses, and the like. Mark Gardiner says that the Anglo-Saxons might have established ports at Hamwic (Southampton), Sandwich, Dover and elsewhere. They are not known to have built any freight handling ports from new, so they were probably re-activated Roman ports. If so, the port gifted to St Denys was associated with saltpans, so it was probably at the mouth of the Brede.

So, the physical evidence suggests that the Brede discharged into the Channel near modern Winchelsea in Romano-British times. It would have had to pass through a breach in the Camber shingle bar. It is on this bar that the Romans would have built their port for transferring the Brede basin’s iron blooms to sea going ships and for salting fish. That port would have dwarfed any others in East Sussex, making it the most likely location of Novus Portus. This place was known as ‘Winchelse’ in Anglo-Saxon times. It is now known as Old Winchelsea.

In summary, our corrected Ptolemaic coordinates suggest that Novus Portus was on the coast between modern Hastings and modern Rye. The Brede estuary was in this range, close to Rye. The Brede basin had four times the iron ore reserves and twice as much salt as the rest of Sussex and Kent combined. Nearly all of it was exported. The enormous volume of natural resources being exported would have needed a freight handling port, and the only place it could have been was at the mouth of the Brede. The mouth of the Brede in Roman times was at a breach in the Camber shingle bar at the place now known as Old Winchelsea. We propose that this was the Roman port of Novus Portus, and that it was guarded by a fortified garrison at modern Winchelsea.

Hæstingaport at Old Winchelsea

Nine hundred years after Geographike Hyphegesis was written, Hæstingaport and Hæstingaceastre are mentioned in the Anglo-Saxon Chronicle. Hæstingaceastre was the second easternmost of King Alfred’s 33 burhs, placing it in East Sussex. They were located, where possible, at the sites of former Roman fortifications. This seems likely with Hæstingaceastre because Old English ‘ceastre’ referred to pre-Anglo-Saxon fortifications, nearly always Roman. Old English ‘port’ usually referred to ports in the modern freight handling sense, different from ‘havens’ and ‘harbours’ which Anglo-Saxons referred to as ‘hæfens’ or ‘herebeorgs’. So, Hæstingaport and Hæstingaceastre refer to a freight handling port and a former Roman fortification in East Sussex. There are obvious parallels with what we propose above for Novus Portus.

The ‘Hæstinga’ part of the names is usually assumed to mean modern Hastings. We explain in our ‘Battle of Hastings Place Names’ paper why it referred instead to the Hastings Peninsula. It only matters here that Hæstingaport and Hæstingaceastre were on the Hastings Peninsula which would apply to both meanings.

Hæstingaport and Hæstingaceastre were adjacent or encompassing. One reason to think so is that is that John of Worcester repeats all the Anglo-Saxon Chronicle entries that mention Hæstingaceastre or Hæstingaport and refers to both Heastinga, as if they are cognates. Another is that Hæstingaceastre had a Grately Code mint. They were only at major population centres and some coastal burhs. Presumably the latter overlooked ports that took payment in foreign coin, to be melted and restamped. It seems that this was so at Hæstingaceastre because some of its coins were stamped ‘Hestingpor’ or similar. A third is that the Anglo-Saxon Chronicle entry for 1049 says that ships and men from Hæstingaceastre captured two of Sweyne’s ships. Roman fortifications were almost never at sea level. The ships would have been moored in a harbour or berthed at a wharf below the ceastre.

Hæstingaceastre was an Alfredian burh, the administrative and military centres for Wessex districts. It held the only Greatly Code mint in East Sussex. It was home to the only burgesses on the Hastings Peninsula. This makes it by far the most important commercial, military and administrative location on the Hastings Peninsula, and it remained so until the Normans constructed the castle at modern Hastings.

It was common practice in medieval times – and still is in much of the world – for ports to physically separate gentlemen from riffraff like sailors and stevedores, and from fetid fish processing plants. We interpret this to mean that Hæstingaport was divided with its mint, garrison, ship owners, tradesmen, and merchants at Hæstingaceastre physically distanced, albeit by perhaps only a few hundred metres, from its docks, warehouses and fisheries.

There are three candidates for a pre-Conquest medieval freight handling port on the Hastings Peninsula, none of which are modern Hastings. Before discussing the viable candidates, we should explain why modern Hastings is not one of them:

1. It was accessed via a long narrow gorge amidst four miles of treacherous sea cliffs. Medieval ports were not built anywhere so dangerous.
2. It had no Anglo-Saxon population and no post-Roman natural resources, so it had no reason to have a port.
3. De Viis Maris, written in the middle of the 12th century, specifically says that there was not a port at modern Hastings. If there was no port a hundred years after the construction and occupation of Hastings Castle, there was not one beforehand when the area was unoccupied.
4. It was at the top of a steep-sided 55m sea cliff, with no easy way to move freight up or down to the valleys below.
5. It had no Roman road, so no easy way to distribute freight to its hinterland.
6. There is no archaeological evidence of medieval occupation at modern Hastings, or of a medieval port or medieval occupation in the valleys below.

The viable Hæstingaport candidates are at the mouth of the Brede estuary, at the mouth of Combe Haven and at the mouth of Pevensey Lagoon. It is known from later records that each had a port in the 12th century, at Old Winchelsea, Bulverhythe and Northeye, respectively.

The best clue about Hæstingaport’s location is that the Rochester Roman road terminated at modern Winchelsea. It was the only Roman road on the Hastings Peninsula. Roman roads almost always terminated at garrison, so there was a Roman garrison at modern Winchelsea, and it was the only Roman garrison on the Hastings Peninsula. The ceastre part of Hæstingaceastre’s name means it was a former Roman fortification, so it was at modern Winchelsea. If Hæstingaceastre was at modern Winchelsea and it was part of Hæstingaport, then Hæstingaport’s docks, warehouses and fisheries were at Old Winchelsea. While we think this is a compelling argument on its own, it is good to check the other evidence about Hæstingaport’s location.

Population provides a useful clue. According to Domesday: Rameslie manor around the Brede estuary had 189 households, 68 of which were burgesses, 35 ploughlands and seven acres of meadow; Filsham, Cortesley and Wilting manor around Combe Haven had a combined 122 households, 57 ploughlands and 57 aces of meadow; Hooe had 71 households, 44 ploughlands and 71 acres of meadow. The presence of a large cohort of burgesses and the far higher ratio of households to farmland in Rameslie implies that it held a major port, and a port in Rameslie is mentioned in the 795 gift to St Denys and in the 1017 gift to Fécamps. Rameslie was beside the River Brede, so the port was at the mouth of the Brede, which means it was at Old Winchelsea.

Hæstingaport’s name makes it sound like the most important port on the Hastings Peninsula. This is one reason we believe that the Normans landed nearby and camped at Hæstingaport before and after the Battle of Hastings. We list thirty or so reasons to think it was at Old and modern Winchelsea in our paper about the Battle of Hastings. Among them, according to De Viis Maris, written in the middle of the 12th century, Old Winchelsea was the only significant port on the Hastings Peninsula. It was where William arrived on his return to England in 1067. It is the only place in East Sussex that Henry II is known to have visited, the occasion being his return to England in 1188. It is where King John commissioned 20 ships to be built for an attack on France. It is where King John stayed in 1213 when he was planning this attack. It is where King Louis took the French army to await rescue after his failed invasion in 1217.

Hæstingaport is the only named Anglo-Saxon port on the Hastings Peninsula and De Viis Maris sys it is the only significant port. This implies it was by far the biggest port. The region did not have enough population to draw significant imports, so Hæstingaport was a natural resources export port. Not iron, because the richest seams were mined out before the Romans left, but salt and timber. Domesday says that the Brede estuary had 100 saltpans, Pevensey Lagoon had 34, Combe Haven had none. Transporting timber on unpaved roads was difficult and slow The Brede estuary was the only place suitable for harvesting timber on the Hastings Peninsula thanks to its paved road and steep banks which allowed timber to be dropped down to the river on log flumes. Indeed, ancient log flumes are still visible in Steephill Wood. The Pipe Rolls entry for 1204 records that Old Winchelsea was the biggest port between London and Southampton, whereas Bulverhythe and Northeye were tiny. Ship Service was an algorithm for providing ships to the navy based on freight traffic volume. According to the 1227 Ship Service records, Old Winchelsea had more than ten times the freight volume of Bulverhythe and Northeye combined. By these measures, Old Winchelsea was by far the biggest Anglo-Saxon port on the Hastings Peninsula which means it was almost certainly the docks for Hæstingaport.

To summarise, Hæstingaport’s name implies it was the biggest and most important port on the Hastings Peninsula. All the evidence indicates those accolades belonged to Old Winchelsea. Hæstingaceastre’s name implies it was the main Anglo-Saxon military and administrative centre in East Sussex. All the evidence indicates it was at modern Winchelsea. We therefore think that Old Winchelsea was the main port in East Sussex from Roman times through to the 13th century when it was destroyed by storms.

Some loose ends

If there was a port at Old Winchelsea and a fortified garrison and a paved Roman road at modern Winchelsea, why were they not used when the Romans built the Saxon Shore Forts in the third century? Instead, the Romans built shore forts 17 miles west at Lympne and 19 miles east at modern Pevensey.

One possibility, as Andrew Pearson pointed out in his 2001 book about the Saxon Shore Forts, is that Notitia Dignitatum is “a notoriously unreliable guide to the occupation of a site”. He explains that many sites listed therein had closed or moved by the time it was written. Perhaps it was omitted by mistake. We suspect an alternative.

The Saxon Shore Forts were built to defend against raids and invasions. Once the Brede’s iron ore ran out, there was nothing left to attract raiders, and it was not on the way to anywhere that invaders might want to go. We guess the idea was that the fortress at Lympne defended the route to Canterbury, while the fortresses at Pevensey defended the route to London. Remember that the Rochester Roman road to Winchelsea did not go to London.

One other loose end concerns the link between λιμην, Lemaris and Limen, the Old English name for the Rother. This is the only evidence supporting many of the Novus Portus candidates listed above. We think it is a distracting coincidence.

The first thing to note is that Ptolemy was writing in Greek but all the Geographike Hyphegesis places in Britain had Latin names. He transliterated most of them into the Greek alphabet, or translated those that had no acknowledged name into Greek. Novus Portus would have been one of the latter. It was never known as καινòς λιμην anywhere other than by Ptolemy. Therefore, Lemaris and Limen are unrelated to καινòς λιμην.

It seems far more likely to us that Lemaris derives from the Latin noun ‘lama’, meaning ‘marsh’ or ‘swamp’. Portus Lemaris would therefore be ‘Marshland Port’, or some such. The Anglo-Saxon name Limen for the River Rother might then have derived from lama or Lemaris, but we do not discount Simon Mansfield’s theory that it comes from the Old English word ‘Læmen’, because it was silty.

Bibliography

Claudius Ptolemy; Geographike Hyphegesis; 150 A.D.

 

Charles Anthon; A system of ancient and mediæval geography for the use of schools and colleges
Edward Wedlake Brayley; A topographical history of Surrey; 1850; Willis
Henry Cleere; Roman harbours in Britain south of Hadrian’s Wall; Roman Shipping and Trade; 1978
Lawrence Echard; The Classical Geographic Dictionary; 1715
Robert Furley; ‘An outline of the history of Romney Marsh’ in Archaeologia Cantiana; 1880
Mark Gardiner; Economy and landscape change in post-Roman and early medieval Sussex 450-1175; The Archaeology of Sussex to AD2000; Heritage Marketing and Publications; 2003
John Horsley; Britannia Romana; 1732; Longman
Rupert Jackson; The Roman Occupation of Britain and its Legacy; 2021
Bernard Leeman; The 1290 Masacre of Jews at Jury’s Gap Romney Marsh; 2015
Andrew Pearson; The Roman Shore Forts : Coastal Defences of Southern Britain; 2002; Tempus Publishing
John Peterson; Le réseau centurié ‘Kent A, Atlas Historique des cadastres d'Europe; 2002; Office for Official Publications of the European Community, Luxembourg
Beale Poste; Britannic researches; or, New facts and rectifications of ancient British history; 1853; J R Smith
Shirley Read; Bygones of Bexhill; 1992; Bexhill Museum
William Somner; A treatise of the Roman ports and forts in Kent; 1660