Roman Road to Winchester


A large water feature was found in excavation in 2012, to the north of the Roman Centuriation estate road or limites excavated in 2011 - where this feature was detected as a depression in an optical exercise (Ref 1). A plan of this depression is given in Ref 1 and its relation to the Road, which was an engineered terrace, lightly metalled with crushed flint, and appeared to be unusually wide at 40'. A similar plan is given in Diagram 2 below, with the 2011 and 2012 trenches. The 2012 Composite Section is shown in Diagram 1 below - a larger landscape version can be printed at the end. Starting at the southern end (9N) we have similar stratigraphy to that beyond the Roman Road to the south - Top Soil and subsoil on two layers of natural clay.

DIAGRAM 1 Composite Section. The Trench Numbers are shown below the Ground Scale, each 4m long.

Then we come to the packed flint feature at 11N. Diagram 2 shows it is much smaller than that found in 2011 - which was assumed to be hard standing at the edge of water. Thus this hard standing did not extend along the edge, but was confined to a small part. From this we can conclude that the unusually wide road we thought we found ending at this packed flint feature was because we happened to be excavating along a lane leading from the estate road to this water feature - the hard standing to enable carts to gain access for carting water elsewhere. In this regard we did not see any of the chalk metalling found in 2011 ending at the parked flint - but in 2012 crushed flint occurred in the subsoil up to the start of the silt at 16N. This is a lot further north than seen in 2011. It could be seen in section along the clay boundary as shown on the composite section. This crushed flint was probably to help animals get the waters edge for drinking.

At 14N - 16N the natural statigraphy has been altered. The upper yellow clay band ends, and from 16N replaced by a Silt layer. If this is silting up of a water feature which went out of use in Roman times then the subsoil above will have been formed since, and its understandable that the crushed flint ends here. At the start of the Silt at 16N the section photo indicates that a ditch has been cut, and so the Red Clay may have been re-deposited from 16 N to 14N increasing the height of Red Clay here - the Yellow Clay having been removed throughout. The section photos may suggest that this clay has been re-deposited. The Silt layer continues out of the trench at 20N. The dotted line (of fine dots) is the likely water surface level. This is taken from the likely relation to the Parked Flint hard standing in the 2011 Composite Section - but it also agrees with the surface of the Silt layer which is also likely to reflect the original water surface at -0.5m.

Over 1 m deep

The trenching starts again over the north edge of the optical depression, Trench 13A, 36N - 40N. Here the clay is rising steeply having been cut down over 1m from the likely water level. The surface of the cut seemed to be studded by small stones. As can be seen from Diagram 2, the edge of the cut suggests we are near the eastern edge of this water feature. The solid line of the Red Clay boundary (with the stones) is that on the east face of Trench 13A - but the bottom of this water feature slopes steeply within Trench 13A - and the form of this clay boundary on the West Face is shown as the dashed line. We probably have not reached the bottom of this feature. One of the stones on the boundary was a likely worked flint, indicated by the arrow.

The fill is a Golden Brown Silt. This raises the question if it is also subsoil - as subsoil takes around 1000 years to form and can approximately date features. The silt in Trench 8A is paler and can be taken as also subsoil, and is consistent with the water feature silting up in or shortly after the Roman period. Further evidence in this regard comes from the section drawings of the South, and West, Faces of Trench 13A. These show Red Clay over the Golden Brown Silt, with Top Soil on the clay. This is not a natural stratigraphy. This indicates that clay has been re-deposited on the Goldern Brown Silt, with Top Soil forming on the clay without sufficient time for subsoil to form - giving a Medieval date for this clay deposit. Indeed the East Face of Trench 13A - as shown on the Composite Section - also does not show natural stratigraphy. The Golden Brown Silt from 37N to beyond 40N might suggest the water feature had a shallow northern edge, perhaps to assist animals drinking - but the feature would then extend beyond the optical depression. While the section drawers have taken the thick layer from 36N - 38N as subsoil, the section photos show a large red content which may be a clay mixture.

Medieval interpretation

Furthermore the projected water surface only just clips the Top Soil in Trench 13A - not the Silt surface as in Trench 8A. This suggests erosion of the downhill edge, which is to be expected, with re-deposited material to build-up this edge again. While the shallow edge in Trench 8A would silt up quickly, the much deeper northern edge would take much longer to silt up. So the evidence is that this water feature was still operating into the medieval period when its northern edge was rebuilt using a variety of materials which could include a soil-clay mix and excavated Golden Brown Silt seen in the East Face of Trench 13A. There is not evidence that this water feature had the silt cleaned out - which may produce different coloured bands of silt. Eventually the northern edge became silted up and the feature was forgotten.


After the excavation Magnetometary was undertaken to check that the depression area had a substantial silt layer with cut down clay. The method used was developed at Roe Downs Farm to quickly check particular blocks of land by noting the swings of the Signal Meter as the detector head past over the land (Ref 2). Four traverses were done, shown as Bold type on Diagram 2. There were 5 blocks of land: (1) & (2) Either side of the E - W base line (on the Roman Estate Road), (3) 8N - 20N (northern coordinates of Trenches 6A - 8A), (4) 20N - 36N (the likely deepening water feature), (5) 36N - 40N (over northern edge of the water feature). Figures less than 10 show lower ground magnetic effect from where the equipment was set up.

The results in Diagram 2 show high magnetic effect over the likely deepening water feature. Over the Roman Road readings are low. Over the start of the silt they are higher (except the most westerly, see later). Over the likely increasing depth of the water feature we have the highest readings recorded (here the common position of the meter needle is also given). Over the northern edge they fall to similar levels as over the southern edge. Soils are more magnetic than clay, so a silt layer building up to a 1m thick can be expected to give much higher signal.

Advanced Roman Dew Pond?

The above interpretation assumes this feature had a water supply which continued after the Romans. On the face of it, this is a most unlikely position for the Romans to have built a vast 20mx20m water storage facility - let alone there being any source of water to fill it. As Ref 1 shows the land is falling slightly to the north, and this continues for c 200m to the south when the summit of the downs is reached. On the other hand, Roman water technology probably exceeded ours. Apart from the well known aqueducts, and city storage, distribution and drainage systems, N Africa was a prime granary - though now it is mainly desert. Only recently have we got any idea how they did it.

I was born on farm with a Dew Pond, and remember it. I went back there recently, and although now filled in, located it. It was in a most unlikely position, at the bottom of a saddle. That is to say, coming up the track from the farm the land rises to a summit and falls away to the valley below. But in the E & W directions the land rises for a while. Yet there was a self sustaining pond there, heavily shrouded with trees. Wikipedia says it is not known how dew ponds work, but they are found near the tops of hills, and are reputed not to dry out even in prolonged droughts - Gilbert White of Selborne studied them with this conclusion (Ref 3). Pears adds that the name comes from the belief that dew deposits at night replenish water used by stock, but drainage of rain water and mist condensing on trees are probably more important (Ref 4). The landowner says there are Dew Ponds in our excavation region, on the sides of hills near the summits. The website gives pictures of old and modern Dew Ponds - generally without trees and often on saddles - with stock around them, saying it is now generally regarded that rain is the main source.

Thus there is substantial evidence that these ponds exist, so it should be possible to identify how they work, and so how the Romans may have advanced them. Dew is doubtful as the necessary trees will take a great deal of water out of the soils - but citing near the tops of hills will cause the air to cool as it is forced up the hillsides, increasing condensation. As Pears says drainage of rainfall is likely to be the main source. In this excavation site, the natural is clay which largely prevents rain passing downwards - instead it must seep down the hillside through the soils or on the surface. If you then dig a pit in the clay it will eventually fill with water. In some strata the rain may be stored in the hill if there is some barrier lower down such as clay. Then if you dig a pit, and suitably line it with clay so as to allow water to enter from up-hill but prevent it leaving down-hill, then it will fill with water. This type may be more sustaining in dry weather.

We can see why these ponds are near the tops of hills. Landowners on the Hog's Back and Chawton Park Wood say in heavy rain it runs down these hillsides as torrents. On the Hog's Back the Roman Road was protected by a large up-hill ditch - otherwise it was liable to be washed away. Citing these Ponds near the summits but where there is sufficient up-hill catchment area to keep them filled prevents such damage in heavy rain.

Roman advance

These modern ponds catch their water from the diameter of the pond times the distance up hill to the summit of the downs. The Romans could have extended this by an order of magnitude by making the catchment device not the pond itself but a suitable ditch which catches the water as does the up hill edge of a Dew Pond - for several hundred meters or more. A project was mounted to see if this water feature and its catchment ditches could be identified on air photos.

Air photos were examined which were used to produce the Roman Road survey maps, Ref 5 & 6. The lightest photocopy of Ref 5 show a darker mark in the area of the optical depression, which becomes pronounced in the contrast enhanced version of Ref 6. The image of these are placed on Diagram 2, the arrows indicate a tendency to get darker.

No other ditches could be seen going from this optical depression in relevant directions - though this lightest photocopy probably provides the northern side ditch of the Roman Road missing in the published survey. In order to detect ditches by our air photo methods we need to know the likely direction they take. Thus consider what directions would the Romans have constructed feeder ditches. The maximum amount of water will be collected when the flow or seepage of water through the soils is at right angles to the direction of the ditch. The Romans would be concerned to minimise the labour of digging such ditches - or more importantly keeping them maintained and free of silt so they continue their collection function. This condition is met when the land has a slope but the ditch's direction of travel is level (unless the land has unusual shapes).

We can thus look on maps and search in directions parallel to height contours. The only relevant direction from our water feature is the Roman estate road itself - where on Ref 7 Map the 125m contour snakes around our Road from the excavation site for near 200m eastwards - just about the distance we failed to find the northern side ditch in preparing the published survey plan. We can thus take it that the Romans used the side ditches of their Roads which they were going to cut anyway - whenever the above conditions occur. The position to store the collected water would be at the end of the level way where the land falls below the level - just where we found our water feature.

The flow of water across the road will be SE to NW, but it is not readily apparent that there is such a slope as it is slight - or that Road is running level over the catchment area - except from our composite sections and looking at modern contour maps. So high level of surveying skills must have been used to detect when the conditions occur to create such a water facility. This is a Centuriation estate where providing such a water source would be valuable - it is a question if this was done on ordinary Roman Roads. There are only a few cases where Centuriation has been found in Britain with estate roads (Ref 8) - it would be worth searching on these where the above conditions may exist. We were fortunate to have an excavation site by such a water storage facility, with a trench along the lane from the limites to the water facility - thus establishing a connection.

Ditch with silt lines

There was a very interesting ditch seen in the 2011 excavation just at the northern edge of the packed flint hard standing, visible on the section photos, with re-cuts and silt lines. This can be seen on the section photos without any special processing, and can be viewed at This ditch is narrow but deep, and coming onto the hard standing could well be where the water collected enters the storage facility. It would indicate we were excavating very near its east edge. This may be the northern side ditch of the road, but from above we would expect the southern ditch to be the water collector. The collection ditch would have to be cut into the clay to collect the water seeping through the soils, but keeping it narrow would reduce evaporation. The collection ditch would have to get deeper as it went west to to provide a flow - but the Romans could have assisted this by picking the run of their collection ditch so it had a slight fall - a few degrees off being at a right angle to the direction of water flow through the soils will not make much difference to the volume of water collected. Careful examination of the Ref 7 Map and its contours suggest this is what the Romans did. The critical issue would be collecting the water from heavy rains - if the flow was not enough the water would overflow the ditch and be lost. Possibly both ditches were used for this problem.

The collection area is likely to be over 40,000 sq m, which is 20 times what one dew pond of 10m diameter would have. The ratios of volume of water stored to catchment area seem to be similar for both. The main advantage of this water facility is that the ratio of surface area to volume is around 1/3 that of a pond - so the loss by evaporation would be much less. If dew or mist has any contribution it may be lower for our water facility.

Optical depression

The results of the 2011 optical depression, and 2012 magnetometary with the recent air photo image are shown in Diagram 2 . They all tend to suggest that the water facility gets deeper as it goes west as well as north, and is not at right angles to the estate road. We know from the excavation that the depth increases as we go north, to over 1m deep at 36N. Yet the air photo image hardly reaches this position in the 2012 trench. The air photo tends to get darker in the direction of the arrows, suggesting increased silt layer. That the feature is not at right angles to the estate road is suggested by the optics and air photo. The magnetometary traverse at 20W has the largest and smallest numbers, which could be as 36N 20W is outside the water facility while say 20N 20W is within it.

The arrangement where the deepest part was in its NW corner say around 35N 15W would mean the water would remain useable as it was used up both for carting and animal drinking. The water would naturally congreate into a smaller area and loss from evaporation would be reduced.

Diagram 2 The 2011 optical depression, with the 2012 trenches over its edges, the magnetometary, and air photo evidence added. The archaeological grid is along the Roman Road (E - W) and right angles to it, the terrace found in 2011 is shown. The magnetometary data along four N - S traverses is in bold type. The arrows indicate how the air photo image becomes darker.


Thanks are due to Mrs Gillian Gray Knight for permission to excavate on her land, and to the 39 man days contributed by the diggers. Richard Whaley


1. Richard Whaley, Centuriation limites were lightly metalled, NEHHAS FAB e News 4 Spring 2012, http//
2. Richard Whaley, Magnetometary on the line from Zig-Zag NEHAS FAB e News 6, Winter 2012 http//
3. .
4. Pears Cyclopaedia, General Information, p543. 62nd edn 1953, A & F Pears
5. OS Air photo 013 71 387 (1971). A further lighter photocopy was made.
6. Air photo from, 2011.
7. OS Explorer Map 132, 1:25000, 2009
8. A Richardson, Trans Cumberland & Westmoreland Arc Soc 86 p71, 1986; S Frere, Britannia XXX1 2000, p350.

Printing Full Size Landscape Copy of the Section Drawing

Click here then check your margins 9mm and set to Landscape [File - Print Set-up - Landscape, Margins 9mm. File - Print..] After Printing, re-set to Portrait and default margins.