Excavations took place at the renowned site of Mucking, Essex, between 1965 and 1978; while a 'Rescue' excavation, taking place in advance of (and often during) quarrying operations, it also aimed at near-total excavation of the archaeological features. The site was directed by Margaret Jones, but her Assistant Director (and husband and site photographer), W.T. (Tom) Jones took the main responsibility for directing the cemetery excavations, co-ordinating much of the initial post-excavation research on them. Along with the well-known Anglo-Saxon settlement and cemeteries (Hamerow 1993; Hirst and Clark 2009), the excavations produced extensive evidence for prehistoric phases of occupation, and for a large Romano-British settlement, complete with four formal cemeteries and evidence for pottery production in the form of kilns and their products (Figure 1). Publication of these aspects of the site is now in progress, following a project hosted at the Cambridge Archaeological Unit, and funded by The British Museum, The Society of Antiquaries of London, English Heritage, the British Academy and the Roman Research Trust (Lucy et al. forthcoming; Evans et al. 2015).
The Romano-British settlement at Mucking exhibited direct settlement and production continuity from the later Iron Age, but saw the laying out of a more formal rectilinear enclosure system around the middle of the 1st century AD. In the 2nd century AD, these more 'formal' parts of the settlement continued to be occupied and developed, but the western enclosure area (in use from the later Iron Age) went out of use. Both of these phases saw contemporary burial, both inhumation and cremation, within distinct cemetery areas. In the 3rd and 4th centuries, the picture is more obscure. Very few cut features can be assigned to the later Roman period; some burials can be seen to date to the 3rd century, particularly the earlier part, with only a handful, at most, dating to the 4th century, but all indications are that the excavated area saw a dramatic decline in settlement intensity, if not actual abandonment after AD 250. A key issue is whether the relatively small amount of distinctly 4th-century material culture actually represents activity of that date, or whether it was associated with the subsequent Anglo-Saxon phase of occupation.
As outlined more fully in Evans et al. forthcoming, our interpretation of the pre-Anglo-Saxon settlement sequence at Mucking has been heavily dependent on the two previous phases of post-excavation: that of the Jones' Mucking Post-Excavation (MPX), which ran from 1978 to 1985, and that of the British Museum/English Heritage team (BM/EH), from 1986 to 1992. From the latter phase, we have drawn heavily on the stratigraphic sequence for the site as depicted in the Site Atlas (Clark 1993), although the original site records have also been extensively used. Fundamental to our work has, however, been the work of MPX, particularly the computer cataloguing of many aspects of the archive, and this is particularly true for the interpretation of the Roman settlement. The Mucking archive is actually relatively easy to interrogate for these purposes. To aid the post-excavation process a microcomputer was acquired by MPX (see Catton et al. 1981) and data-input files written to catalogue small finds and to produce rudimentary, but effective, site distribution analyses and plots of some finds. Small finds and feature categories (graves, structures, drawing numbers etc.) were entered onto separate databases with a unique serial number allocated for each entry within an individual dataset. All data entries contain information on co-ordinates, level and notebook references. Not all catalogues were started or completed, but a substantial amount of information was recorded.
Data was stored on 8in. floppy disks in 1983 and retained within the archive, although the decision was taken by the BM/EH team not to employ or migrate these data during the second post-excavation phase. This decision does not seem to have been properly justified, beyond citing 'problems inherent in the record' (Clark 1993, 2). Fortunately (and rather to our astonishment) the data contained within these files proved both intact and usable, after extraction by a specialist company. The data input programmes (known as 'Question Source Files' were well thought through and tightly structured, meaning that consecutive strings of data, each representing an individual record, could be reassembled into Excel spreadsheets, albeit with some considerable investment of time and effort. Thus, although the notebook recording at first sight appears rather chaotic (the description of a single feature is often spread over several parts of one or more notebooks), the computerised archive provides the 'key' by which it can be unlocked. Datasheets rescued from the 1983 disks include the notebook index (feature numbers and notes recorded against northing, easting, notebook number and page), and the plan and section index (ditto). After the relevant notebook pages and plan and section numbers have been identified, the relevant drawings and notes can be examined on the microfiche that exists of them all.
It is the finds recording, however, that has proved fundamental. While the Jones' programme of computer cataloguing has been much criticised (e.g. Clark 1993; Moffett 1989), rescue and re-examination of the datasets has prompted a re-evaluation (see Evans et al. 2015 for more detail). With relevance to the Roman settlement interpretation, the following datasets were employed: Romano-British pottery (with separate datasets for samian and mortaria); worked stone; iron; animal bone; fired clay and ceramic building material. No computer records existed for glass, copper alloy, brooches or coins, and these datasets were created in 2008, with specialist reports commissioned. All MPX datasets followed the same format in terms of contextual information: records always included a unique serial number, northing, easting, level, depth, feature type, feature number (where applicable), context description, and then relevant details of the material being recorded. Given the sheer quantity of finds generated by the Mucking excavations, some of these are immense: the Romano-British pottery file, for example, comprises over 45,000 rows of data relating to over 140,000 sherds, with relevant data recorded in 33 columns, including form, fabric, decoration, treatment and rim and base EVEs. While MPX was criticised for the detail in which the finds had been recorded (Clark 1993, 13), it has in reality proved invaluable when interrogating the various finds categories for significant patterning and associations, and it was during this process that some unusual patterns were noted within the later Romano-British pottery distributions.