David Roffe

To the manor born: freemen and sokemen revisited [slide 1]

The Haskins Society Conference, 2 November 2012


For the PowerPoint presentation that accompanies this lecture, please contact David Roffe


I am delighted to be here at the Haskins Society conference and I am most grateful to the programme committee for inviting me to speak. It is, of course, a culture shock for me to come here, hailing as I do from rural Lincolnshire. Wow! Is Boston the big city? We don’t have a subway in Lincoln or, indeed, anywhere else. Goodness, in most of the county we don’t even have buses. It’s not, though, as if I am completely unprepared. After all, we too have high rises. In fact we’ve had them somewhat longer than you, for at least 500 years [slide 2]. I bring you greetings from that other Boston.

            As you will have divined, my subject today is the manor in the eleventh century. I must warn you at the outset, however, that my conclusions are very tentative. They come from the early stages of a major new project in medieval studies. In September 2011 a conference was held at The National Archives in Kew in south-west London, to take stock of Domesday studies since the last audit in 2000. Much had happened in the intervening years. Challenges had been made to all the accepted wisdoms of Domesday studies – the nature of the business in 1086, its relationship to the Domesday texts, the referents of data, and so on. It seemed like a good idea to see whether any new consensus was emerging. All round it was an excellent day. There were seven first-class papers and the stimulating discussion that resulted continued in the pub well into the evening. We all learnt a great deal. I can’t say that we reached any great agreement on substantive matters. What did emerge, though, was the urgent need for a new edition of Domesday Book and all the texts associated with it. An ad hoc committee was subsequently formed, and the Domesday Texts Project came into being.

            Why a new edition? It is generally agreed that the Farley transcript [slide 3] of Domesday Book is extremely accurate, in fact a wonder of scholarship today, let alone the late eighteenth century when it was published. Indeed, there is something of a fan club out there that resents any suggestion that it is inadequate. But inadequate it is. With benefit of the high resolution version of the Digital Domesday [slide 4] – in my view far superior to the printed facsimile – it is now clear that Farley contains many more errors of transcription than hitherto appreciated. More importantly, however, it is not a critical edition that meets the demands of modern scholarship.

            Here’s a Domesday entry as you never seen one before [slide 5a]. My God, if you are lucky, you’ll never have to look at it in this form ever again. The passage in question is the description of the king’s manor of Merton in Surrey. Phew! That’s better [slide 5b]. It is marked up for the semantic web in the TEI version of xml. Extensible Markup Language is a technology for encoding documents in a form that is machine readable but also comprehensible by mere mortals; the Text Encoding Initiative provides a standardized framework for representing historical texts amongst others. What sorts of new information, then, is represented? Well, we have the full range of scribal history, additions, deletions, changes of hand etc, in fact all that you would expect in a modern edition. Here, for example, [slide 5c] we have comes interlined by the main scribe. So, the code [slide 5d] translates as ‘addition by hand 1’, the main scribe, ‘period 1’, that is during the course of initial drafting, ‘type interlinear’.

           This much is surely a minimum. You might find the coding of contraction and suspension marks more surprising. Here, for example, I have recorded the missing ‘et’ in tenet [slide 5f]. Nevertheless, it is timely. At the conference last year John Palmer read a wonderful paper in which he demonstrated that there were patterns to the distribution of different diacritics. He thereby opened up a whole new category of evidence for scribal sources and methods. Concentration on such minutiae is invaluable. Just recently I have realized, for example, that the up-side-down semicolon [slide 6a] – the only name I can find for it is the Arabic fasila manquta – has a very specific use: it indicates the omission of a word or a number of words. Its implications for expanding the text and translating are considerable. Silva x porc’ , for example, is usually expanded to silva x porcis and rendered as ‘woodland for so many pigs’. However, the fasila manquta often follows silva [slide 6b] and shows, I think, that we should understand something else like silva: x porci de pasnagio, that is, ‘so many pigs from pannage dues from the wood’. What too often appears in modern translations as a measure of woodland becomes the measure of a pig render that it was.

            Entirely superfluous, you might be forgiven for thinking, is the further encoding of the form of initial capital letters. Be that as it may, this matter brings us more closely to the ostensible subject of this paper. Many years ago I drew attention to a system of encoding that was used by the GDB scribe himself. There really is nothing new under the sun. An underlying assumption in Circuit VI, the northern counties, was that every parcel of land belonged to a manor and could be characterized as either held by the lord or subordinate to him. Marginal letters – M for ‘manor’, B for ‘berewick’, and S for ‘sokeland’ emerged as his main markers. However, from the very start of his work he also used initial letters to signal status. A square I [slide 7a] indicated a manorial centre, echoed in the use of capitals in the place-name, and a rustic I subordinate entries [slide 7b]. Changes from one form to another demonstrates that he took these conventions very seriously.

            So much for Circuit VI. It is now generally agreed that its counties were the first to be written. After enrolling Huntingdonshire, the last in the series, the scribe then turned his attention to Circuit III, almost certainly beginning with Cambridgeshire. From the start, he jettisoned marginal B and S, but initially retained his marginal M. The tight manorial structures of the North, by contrast, are not apparent in the text at all apart from the odd notice of one parcel of land belonging to a manor elsewhere. What is noticeable, however, is that he continued to rigorously distinguish initial letters of different types of entry.

            I shall illustrate with the Cambridgeshire lands of Count Alan. Thirteen manors in all are recorded which were held in 1066 by Edeva the Fair or her men. Twelve of the entries commence with a large square letter [slide 8a], either I for ‘in’ or ‘Ipse’ or the first letter of the place-name. The exception, Landbeach, probably attests an uncharacteristic lapse in concentration. The scribe appears to have thought that the holding was simply land, for he omitted marginal M and wrote the name in lower case letters, only subsequently to discover that Landbeach was a manor. All the remaining entries [slide 8b], identified simply as terra, commence with a rustic I. Of these, the majority, 56 out of 69, relate to lands held in 1066 by thegns or sokemen of one kind or another who were in all but two cases men of Edeva the Fair. They are mapped here [slide 9a]. The circles represent those lands that could not be withdrawn and diamonds those that could. Interspersed with their holdings are a further thirteen parcels of land [slide 9b], indicated by blue circles, held by Edeva the Fair herself.

            The usage looks very deliberate, as it was demonstrably so in the counties he had just enrolled before he began his work on the Cambridgeshire folios. Social and economic structure cannot have been fundamentally different in the two areas. The Cambridgeshire thegns and sokemen must have been very close in status to the holders of sokelands in Circuit VI, and the lands of Edeva look very like berewicks. Indeed, [slide 9c] Toft and Haslingfield are specifically said to belong to her manor of Swavesey. Do we have here, then, evidence that the Domesday scribe perceived manors in the tenurial landscape of pre-Conquest East Anglia as he did in the North?

            The possibility is of some interest, not the least because the historiography has predominantly argued for post-Conquest manorialization and then on an ad hoc basis. The grounds for such a view have always appeared strong. In entry after entry we are told that in 1066 freemen and sokemen ‘could go with their lands’ and the like, but by 1086 they were attached to manors or their land was held by Normans. However, things are rarely as they seem in Domesday Book. A lively twelfth-century land market in Lincolnshire was not incompatible with a manorial structure either TRE or TRW. Nor, by the same token, is freedom of alienation of terra, ‘land’, with similar structures in eastern England. I think we now have to accept that there was no simple equation of lordship and land. As Domesday tells us time and again, commendation was a personal bond. Norman lords might want to see it otherwise in 1086, but the principle was accepted that it did not confer sound title to land.

            Soke was a different matter. A slippery concept it certainly was. It seems to me that it could be used of almost any relationship in which dues passed hands. But that is not to say that it was simply an insignificant archaism as has recently been argued. In the north the soke that equated with consuetudines, or simply terra, trumped all other claims on land bar an explicit grant of the king. The clamores of Huntingdonshire, Lincolnshire, and Yorkshire reveal that tenure in 1086 was firmly grounded in such rights. Similar arguments have been made for Norfolk and now the same mechanisms have been found in Cambridgeshire. In a paper read at our conference last year, Ann Williams showed that the division of the spoils in Armingford Hundred after 1066 were determined by soke.

            We must conclude, I think, that much of what went for lordship in 1086 consisted less of land than rights over land. Freedom of alienation, then, was not incompatible with a tight ‘manorial’ structure. I use inverted commas here because, of course, I am not referring to the later medieval institution. That manors in the Circuit III counties had berewicks and sokelands attached to them, beyond the few explicitly noticed, seems intrinsically likely. The structure is found in Huntingdonshire and a straw in the wind is the manor of Fen Stanton in that county which extended into Papworth hundred in Cambridgeshire [slide 10a]. Papworth itself is said to belong to it and Boxworth was sokeland. It is difficult to escape the conclusions that further lands in Elsworth, Conington, Swavesey, and Fen Drayton [slide 10b], held by the lord or his sokemen, were also members.

            I might add that all of these lands are introduced by a rustic I [slide 10c]. So, does the device hint at manors in Count Alan’s fee? The distribution of the form is certainly not incompatible with such a conclusion. The clustering of the lands around the manor of Burrough Green [slide 11a] in Radfield hundred is particularly suggestive. That all the sokemen in question could not depart with their lands must surely imply a single nexus of dues and services in Burrough itself. Equally, the free sokelands in the north and west of Armingford hundred look as if they constitute a soke centred on Bassingbourn [slide 11b].

            Patterns of tenure in the rest of the hundred corroborate the impression. The distribution of sokeland belonging to Edeva is echoed in the holdings of Earl Algar, Ely, Goda, and vestigially others [slide 12]. Note how even the multiple entries for Croydon and Meldreth find their counterparts in other fees. I’m afraid that the complexity of the pattern defied my PowerPoint skills, but the same phenomenon is more clearly illustrated in Papworth hundred [slide 13]. There the inlands and sokelands of Edeva’s manors of Papworth and Swavesey interlock with those of the Fen Stanton estate that we have already identified and lands associated with King Edward’s manor of Caxton.

            Structures of this kind are regularly found in the north where they attest a peculiar mechanism of estate formation. In Lincolnshire and Yorkshire large sokes were divided element by element to create many of the manors that we see in Domesday Book. The process is documented in a number of places. In Cambridgeshire, I would suggest, we have exactly the same process. Patterns of tenure of this kind cannot easily be interpreted in terms of the post-Conquest kleptomania that has often been postulated for the county. Nor are they the invention of the GDB scribe. They surely must point to the existence of manors of the kind with which we are familiar elsewhere in the Danelaw. 

            The evidence is, I think, compelling. But I’m sure that you are all saying to yourself ‘Hang on! This is all very well, but why do we not see these putative structures in earlier and later sources?’ It’s a fair question. However, it is one that might as easily be asked of northern society. In the early stages of the Domesday inquest in Circuit VI the manor hardly figures at all. The Yorkshire Summary is largely concerned with the vill and the Descriptio Terrarum of Peterborough with demesne and sokeland within it. Details of manorial structure seem to have been supplied from other sources to produce the text that we have in Domesday Book. Thereafter, beyond the large sokes, its manors with attached inlands and sokeland disappear. This does not mean that they were ephemeral. Domesday attests that the manor, whatever it was called, was a significant features of the tenurial landscape before the Conquest. The Norman Conquest seems to mark a stage in its transformation. I have argued elsewhere, however, that the decisive impetus to change was the settlement in the aftermath of the Domesday inquest itself. The Oath of Salisbury in early August 1086 created new rights in land that in the course of time led to the territorialization of lordship.

            We should not be surprised, then, that extensive manors do not appear in either ICC or in twelfth-century sources for East Anglia. Why not in Domesday, though? The answer to that question probably lies in the scribe’s ambivalence to the institution. There are passages in both Yorkshire and Lincolnshire in which the account proceeds geographically. The Isle of Axholme is a prime example [slide 14]. There the texts looks very like the Circuit III folios apart from the ritual addition of marginal B and S. The scribe seems to have been feeling his way. The manor of Yorkshire is a subtly different creature from that of Lincolnshire  After the compilation of Huntingdonshire, the last county in Circuit VI, his interest began to wane. Marginal M was used reasonably consistently in Cambridgeshire, Bedfordshire, and Hertfordshire, but thereafter disappears. The scribe refined his programme in the course of compilation and in the process realized that the manor was not an essential part of it.

            I conclude that the freeman and sokemen of East Anglia were as much part of manors, albeit as an evolving institution, as were their peers in the Northern Danelaw. Their values in both 1066 and 1086 are, I would further argue, a concrete witness to the fact. In the north sokemen are not separately valued. Their dues are engrossed in the figure appended to the demesne. We have no indication of the recipient of the values elsewhere in Domesday Book. However, they were clearly sums that were paid out [slide 15]. The hidagium that the sokemen of Bury St Edmunds rendered in the twelfth century in acquitting their dues to the abbot equates with Domesday values. The valuit and valet figures of sokemen and freemen were not an eleventh-century quantity surveyor’s estimate of productive worth. They were dues paid to a lord and they must have been rendered at his hall. It looks as if sokemen were indeed to the manor born.

            Well, we’ve come a very long way from the form of initial capital letters and the Domesday Texts Project. I hope that the excursus has shown more broadly that the new edition will pay dividends in terms of novel avenues of research. The electronic texts will, of course, facilitate many analyses that have hitherto been difficult or impossible. If the study of the diplomatic of the Domesday text has advanced by leaps and bounds in the last few years, just think what we shall be able to do with a machine-readable text. The ability to cross reference one source to another speaks for itself. So does the ability to switch from a Domesday order to a hundredal one at the touch of a virtual button. The tagging of persons, places, and entries will also enable integration with other online resources. With current database technology there is no reason why the arrays produced by John Palmer and his team should not be accessible through hyperlinkage. The COEL database, soon be put online, will be integrated into the system and it is hoped that functionality can be shared with PASE.

            No doubt you are agog to see a sample of the output. I’m afraid that I cannot demonstrate all the whistles and bells. A basic technology for representing xml files is xstl – extensible stylesheet transformation language. I’m afraid that the procedures I have written so far are too buggy. What I really mean is that they wont work for me since my programming skills are inadequate at the moment. The best I can do is render with css – cascading stylesheets [slide]. Here I can only do basic transformations. For example, initial letters are flagged but I cannot represent their differing forms (there are five in total). Again, additions are indicated but not the periodization. Css does not allow me to render element attributes, or at least not easily. This is the best I can do today.   

            The potential is, I hope you will agree, awesome. The Domesday Texts Project is a long term collaborative enterprise. How long it will take and the progress it will make will depend on the flow of funds. A start has already been made. Surrey has been marked up and Frank Thorn has produced a new open source translation to go with it. We want no pay-walls or other restrictions on free use. As you might have already gathered, work is also underway on Cambridgeshire. The GDB folios are more or less processed but the transcription of ICC lags behind. So much to do. Wish us luck [side 17 – final].


©David Roffe 2012