A large curving dyke with western ditch enclosing the Late Iron Age Settlement at Sheepen (See also MON820 & 728). It is generally thought to be one of the earliest parts of the dyke sequence along with Shrub End Dyke and Heath Farm Dyke. It has been tentatively dated to the Sheepen 1 period (AD5-43). There is evidence for revetment and turf facing of the rampart. Allowed to silt up during Roman occupation.

Excavations in August 1932 on Lexden Dyke Middle. The excavations were undertaken by Mr. Poulter and Miss Cruso under the supervision of C.F.C. Hawkes. Several trenches were opened in an attempt to examine the ramparts construction and prove the existence or non-existence of entranceways through it. <1>

Site 1
Two sections were dug, one across ground where the rampart ought to have stood. This uncovered the ramparts gravel base which had apparently been quarried out. The second was cut through a 9m length of the rampart's base. This showed that the rampart was laid over level natural sand, which had been cleared of turf prior to the banks construction. The rampart was made of gravel with no visible core construction. No trace of revetment or turf was found, but Hawkes believed that it may well of had turf cover. This rampart stood on a natural scarp, which had been steepened by the builders, with a counterscarp dug at the bottom. Evidence of turf thrown down by the ramparts destroyers was found in the counterscarp fill.

A single datable object was found embedded in the top of the last tip of gravel 0.5m down from the surface of the present day slope. This was a Roman bronze plate brooch dated to cAD100. This could indicate that the rampart was slighted before this date, and the brooch dropped in still loose gravel.

Site 2
A narrow trench was placed across the entrance of the dyke. It appeared that the entrance was not chosen because of the narrow gully but because it was where the approach from the valley would be easiest. A spread of gravel metalling appeared to have been laid west of the entrance but this was thought to be no older than the park itself.

Almost 30 fragments of pottery, a fragment of tile, a scrap of iron, part of a Roman looking file and a Roman bronze coin were recovered. Worn and indeterminate Roman pottery was found in the gravely soil which overlaid the metalling mentioned above. The rest was from the entrance itself, some Roman some pre-Roman found in the overlying silty soil.

Two sherds were from the rampart, presumably dropped during the ramparts construction. One from the shoulder of a native coarse ware jar of Cam Form 263, showing a band of fingertip impressions and datable to period one at Sheepen. The other sherd was from a native coarse ware jar Form 264. The position of the Roman material was consistent with initial silting up of the entrance during the Roman period (the coin was dated to Postumus AD259-377)

In the silt a few centimetres above the destroyed northern ramparts base were five pieces of pre-Roman coarse ware, one early Roman sherd, one rim of Gallo-Roman terra-nigra platter of Cam Form 4a, which is usually found in Britain pre-conquest. This does not help give the destruction a very close date but points towards a date in the late 1st century.

Site 3
A 30m break in the dyke was investigated with an L shaped trench. This showed that the break was decided by the natural gully, across which the rampart had run. The trench revealed that the natural sand and gravel had again been stripped of turf. The gravel surface was flecked with charcoal indicating that growth on it may have been cleared by burning. Within the gully itself there appeared to be a clay component to the rampart which Hawkes interpreted as the use of extracted clay to solidify the rampart in an area that was bare of turf.

A 450mm deep pit was found cut into the rampart base, containing charcoal and scraps of incinerated bone, possibly a child burial. May have been a 'foundation deposit' or Lexden cemetery.

Site 4
A complete section was cut through the bank and ditch where it was best preserved. The natural was firmly compacted sand over which the gravel rampart stood to a height of 3m. There was no turf line, suggesting that the turf was stripped by the dyke builders. The bottom of the ditch was 8.3m below the crest of the rampart. The silt in the ditch varied between 2.1m and 380mm in depth, this was mostly sandy-gravel and brown mould. The ditches profile was V-shaped, with a gully in the base 760mm deep. A cylindrical post hole was found 6m in diameter, standing close to the lip of the ditch. Hawkes calculates that the hole would have been 1.2m deep, taking into account the slumping of the bank.

The post hole is taken to be evidence for a timber revetment along the ramparts front, directly behind the post hole the rampart was found to be made up of turf, this was pale grey and cut up into sods, laid in level courses. Hawkes believes that the turf was used to retain the gravely-sandy material behind the revetment. This sat over a core of pebbles without much sand built up as a solid bank 2.4m high. To the back of the rampart a slot was found cut into the sand 380mm across and 530mm deep. This could have been a rear fence but appears too slender a feature to have helped to contain the gravel rampart.

The ditch would have measured 13.4m across and 4.6m deep in its original form. The rampart could have stood up to 4m tall and had a width of 21.3m. No dating evidence was found. There may have been a breastwork, but the monuments crest was too eroded to tell. The turf could have been used in sections of dyke close to entrances so that the rampart could withstand sentries without slumping. <1>