Models of Authority

The Most Important Textual Representation of Royal Authority on Parchment 1100–1250?

Feature Article 3: Dauvit Broun

It is important at the outset of the project to recognise that charters were not the only way in which royal authority was represented in text on parchment (or vellum).[1] There were forms explicitly about the kingship itself. The simplest example would be a listing of kings with reign-lengths, showing the current king to be the latest in a long succession. In the period 1100–1250, however, this was not straightforward for the Scottish kingship. What seems to have been the most common list began only with Cinaed mac Ailpín (d.858).[2] There was no such chronological hesitation with the royal genealogy, often reaching back to Noah or even to Adam ‘son of the living God’.[3] When an attempt was eventually made (sometime between 1165 and 1184) to extend the king-list back before Cinaed mac Ailpín, it was constructed in order to ensure that the genealogy could be followed through the king-list; the architect of this list went so far as to move four eight-century kings in order to make this possible.[4] It was the genealogy that stood centre stage in defining the kingship’s ancient past.[5]

The royal genealogy as the main textual expression of royalty

This is not the main reason for regarding the genealogy as the most significant textual exemplification of the kingship of the Scots. In the earliest detailed account of the inauguration of a king of Scots—an account of the enthronement of the seven-year-old Alexander III at Scone on 13 July 1249—we are told that, at the end of the ceremony,

… a certain highland Scot, kneeling suddenly before the throne, greeted the king in the mother tongue, bowing his head, saying: Bennachd Dé, rí Albanach, Alexanndar mac Alexanndair meic Uilleim meic Énri meic Dauid, and by proclaiming in this way read the genealogy of the kings of Scots to the end.[6]

‘Blessings of God, king of Scots, Alexander son of Alexander …’ announced the Gaelic-speaker who recited the genealogy.

It should be stressed that this performance of the genealogy on such a crucial occasion was not a feat of memory. We are told that it was read out.[7] In the depiction of the royal inauguration ceremony on the seal of Scone Abbey a figure holding what looks like a scroll has been identified.[8]

The seal and the account quoted above were evidently independent of each other.[9] It is safe to suppose, therefore, that in the ritual it was emphasised that the genealogy existed as text on parchment. Although it cannot be demonstrated that it was read out at earlier inaugurations, there is little doubt that it had been pivotal to the portrayal of Scottish kingship for more than three centuries. When a royal dynasty in the male line was first established with the succession of Cinaed mac Ailpín’s grandsons around 890 and 900, a full account of the king’s ancestry from Cinaed mac Ailpín, and of Cinaed’s own ancestry in the male line through a series of definitive ancestors, would thereafter have been essential.[10]

The language of the royal genealogy

It will come as no surprise that the scroll read out at Alexander III’s inauguration does not survive. What about the text itself? Can other copies of the genealogy datable to the period 1100–1250 be traced? According to the account quoted above, the version used in the ceremony would have been in Gaelic. The latest medieval king to appear at the head of a Gaelic text of the genealogy is David I (1124–1153).[11] But this is probably to regard ‘Gaelic’ in too restrictive a way as referring to a text wholly in Gaelic using medieval Gaelic spelling conventions. In an important sense the genealogy was inescapably Gaelic simply by virtue of consisting almost entirely of Gaelic names. From David I’s father, Mael Coluim III back to Gomer, grandson of Noah, there were 125 generations of Gaelic names or names in Gaelic form. The matrix (or ‘host’) language of the text might be in Latin (using filius rather than mac to link the names), but the text itself was Gaelic. There is, indeed, a ‘Latin’ copy of the genealogy in a fourteenth-century manuscript headed by William the Lion (1165–1214) which can readily be recognised as originally a text in which the names were rendered according to medieval Gaelic conventions.[12]

It must be suspected that, by 1249, very few if any of those most closely associated with the Scottish kingship would have been able to read a text in Gaelic orthography. The genealogy, nonetheless, remained at the heart of the most developed literate conceptions of Scottish royalty. When, by the end of Alexander III’s reign, a history of the Scots as a sovereign people from ancient origins to the present day had taken full literary shape, its chronological spine was the royal genealogy.[13] This was possible because, sometime (probably) in the mid-twelfth century, an attempt had been made to convert the spelling of the names into a form that would be accessible to readers who were not familiar with medieval Gaelic orthography.

A new rendering of the royal genealogy

Although the ‘conversion’ to non-Gaelic spelling was not achieved systematically, there was a willingness to reflect actual pronunciation as well as to use letters that were not part of standard Gaelic orthography (k and w). For example, the vocalisation of the velar fricative (pronounced like ‘ch’ in Scots but further back in the mouth), represented in medieval Gaelic spelling by g (as in Éogain and Rogein), is evident in the rendering of these names as Ewein (line 44) or Owan (line 112) and Rowein (line 52). The epenthesis (or ‘extra’ vowel) between n and g in medieval Gaelic Óengusa is found in Enegussa (lines 27 and 59, but not in Engussa in line 25). The palatalised velar fricative represented by g in medieval Gaelic spelling is frequently omitted, again reflecting actual pronunciation (for example Fir Almaig is rendered Firalmai in line 63, and Tigernaig as Thiernai in line 92). Some Gaelic spelling conventions, however, have been retained, or occasionally given an alternative letter. For example the bilabial fricative—breathing through the lips together—which is spelt with m in medieval Gaelic (mh in modern Gaelic) is rendered either with m or u, while the labio-dental fricative—breathing through bottom lip and top teeth together—which is spelt with b in medieval Gaelic (bh in modern Gaelic) is rendered either with b or f. It is obvious that the scholar who created this was not only able to read the genealogy in Gaelic orthography but was a Gaelic speaker who was familiar with its spelling conventions. The earliest witness to this new version is found not in a Scottish text but in the Imagines Historiarum of Ralph of Diss, dean of St Paul’s Cathedral, London, from 1180 until his death shortly after 1200.[14] The earliest extant manuscript of this recalibrated version of the genealogy is in London, Lambeth Palace MS 8, Ralph’s own manuscript of his historical work which he maintained as a kind of final file copy from about 1186 to 1199.[15]

The survival of the royal genealogy

This new version of the genealogy was immensely successful in one crucial respect. The names cannot have been familiar to any of the scribes whose efforts to copy the genealogy we can see today. This includes not only those who copied the extant manuscripts of Ralph of Diss’s work, but also later medieval scribes of Scottish texts which incorporated this version of the genealogy (such as John of Fordun’s Chronicle of the mid-1380s which survives in manuscripts dating from the 1440s to the early sixteenth century, or Bower’s Scotichronicon, composed in the 1440s). Nevertheless the names remained remarkably stable as they passed from one copyist to another (although epithets tended to be dropped): the great majority are readily recognisable in the latest copies. This contrasts with the fate of a version in Gaelic orthography that was transmitted directly through scribes illiterate in Gaelic. This is found from Fergus son of Erc to Noah in a history written in Scots verse by Andrew of Wyntoun, prior of St Serf’s Isle, Loch Leven (a cell of St Andrews Priory) sometime between 1408 and 1424.[16] It also appears independently in the commonplace book of James Gray, secretary of two archbishops of St Andrews in the late fifteenth and beginning of the sixteenth century.[17] It may be suspected that both are derived ultimately from a Gaelic copy that was kept at St Andrews.[18] Be this as it may, the names have suffered significantly worse decay than the version converted to a non-Gaelic spelling, with some so far gone that they can only be identified through comparison with other texts. Scribes simply could not cope.

The contrast between the fate of these names in manuscripts of these versions can be appreciated in the table below, which gives a small section of the genealogy (lines 34–43). The first column gives the names as attested in medieval Irish manuscripts or in the originally Gaelic text headed by William the Lion (1165–1214) that survives in a fourteenth-century manuscript (mentioned above); the second column are the readings in a manuscript of Wyntoun’s history (the ‘decayed’ version derived from a text in Gaelic orthography); the third and fourth columns are the readings from manuscripts of the version with non-Gaelic spelling: Lambeth Palace MS 8 (Ralph of Diss’s own ‘file copy’ of his works), and a manuscript of Fordun’s chronicle (given the siglum ‘E’ or ‘FE’ by scholars: British Library MS Harley 4764). Both ‘E’ and the Wyntoun manuscript (British Library MS Royal 17 D xx, referred to as ‘R’) are datable to the third quarter of the fifteenth century..[19]

The royal genealogy’s survival in a form that was transmitted with its names largely intact is significant for our understanding of textual representations of royalty in a Scottish context. It reminds us that there was a dimension of Scottish kingship as text on parchment that had its origins before charters began to be produced by scribes in Scotland in the twelfth century. It also highlights that this aspect of Scottish kingship continued to be cherished through the period of the project. With its spelling adapted for non-Gaelic readers, it was equipped to serve as the most explicit and potent representation of Scottish royal authority in text not only when a king was inaugurated, but at any time. The kingdom was both fully part of a wider world of textual representations of authority on parchment in charters, and also—through the genealogy—a unique textual manifestation of royal authority.

The integrity of the text of the royal genealogy in manuscripts of Ralph of Diss’s historical works

There is some vivid evidence for how effective the respelling of the names in the genealogy was in allowing its Gaelic names to be transmitted as coherent and comprehensible entities in a thoroughly non-Gaelic context. This is provided by the manuscript copies of Ralph of Diss’s work made by scribes in the south of England.

The point here is not that the names have been copied with exemplary precision, except for some occasional and readily understandable misreadings. What is striking is that there were particular phonemes (mainly fricatives and possibly a couple of others) that the scribes felt they had some flexibility in rendering. This was not random. It suggests instead that they were able to engage with the task of copying this idiosyncratic material without regarding it as forbidding and alien. Presumably they regarded it as a series of vernacular words written in a way that they could hear in their heads or say aloud. Because of this they felt free to deploy their own spellings for phonemes, especially those they were not accustomed to use in Latin but were familiar with in their own vernacular: English. This suggests that the reason this version of the royal genealogy was so relatively stable in a non-Gaelic context was because the scholar who originally created it succeeded in rendering the names in a way that could be read and understood phonemically by those who knew Latin and English but not Gaelic. It is because of this that the text itself was able to retain its integrity more effectively than if it had been transmitted into a non-Gaelic context with its medieval Gaelic spelling unchanged.

The remainder of the Feature will present the data on which this assessment is based. After listing the early manuscripts of Diss’s work that contain copies of the royal genealogy, a summary of the main patterns in the variants will be given before concluding with a new edition of the royal genealogy in these manuscripts. (It has only previously been published as part of William Stubbs’ edition of Ralph of Diss’s historical works.[21]) The particular kind of variability that is most common in the readings of the names represents a special challenge for an edition. This will be met by devising a new editorial methodology which places this variability centre stage, rather than trying to recreate the forms of names in a putative lost exemplar. It might, indeed, be expected that the exemplar was none other than Ralph of Diss’s own ‘file copy’, Lambeth Palace MS 8. There are reasons, however, to hesitate about assuming that this was the case (as will become apparent).

Manuscripts of Ralph of Diss’s historical works

Four manuscripts containing at least some of Diss’s historical works have been identified from before Ralph died or resigned as dean in or shortly after 1199. Three contain the genealogy of the king of Scots headed by William I.[22] These are:

L: London, Lambeth Palace Library 8, a manuscript of 161 folios;

A: London, British Library Additional 40007, a manuscript of 43 folios;

C: London, British Library Cotton Claudius E. III, fos 3–158. 

L and C were probably written at the scriptorium of St Paul’s Cathedral. A, however, appears to have been produced by a professional scribe: it was evidently presented to William Longchamp, bishop of Ely and royal chancellor.

Some manuscripts were produced shortly after Diss’s death:

R: British Library, Royal 13 E. VI (hereafter R), which was copied directly from L after Imagines Historiarum in L had reached its fullest extent, ending with the coronation of King John of England in 1199. Its earliest owner is St Alban’s Abbey.

F: British Library Cotton Faustina A. VIII, fos 40-178 (datable to sometime after 15 May 1208 and no later than early 1209).[23] Its earliest owner is Southwark Priory.

A final manuscript included in the stemma below is:

T: British Library, Cotton Tiberius A. IX, fos 2-48, from Oseney Abbey datable to the mid-thirteenth century; it is closely related to A and F.

There are two other manuscripts, Dublin, Trinity College 503, and BL Cotton Otto D. VII, which Julian Harrison has dated to the thirteenth century, no earlier than 1244.[24] These have not been included in the edition, or had their variants discussed.

Summary of main patterns of variants

On each copying of the genealogy there was always likely to be a limited degree of attrition. This can readily be appreciated if we examine the case of R, whose scribe evidently aimed to produce a faithful copy of L: 

R’s reading are:

Line 26: Fethelmech (L has Fechelmech)

Line 28: Fethelmethe (L has Fethelmeth)

Line 38: Lugtiag (L has Lugthag)

Line 39: Crumgring (L has Crungring)

Line 40: Dare Dromor (L has Daredromor)

Line 75: Bobgai (L has Bolgai)

Line 79: Glachas (L has Glachs)

Line 83: Dein (L has Dem)

Most are straightforward, such as t for c and different readings of minims (which in L’s Dem are ambiguous). The b for l in Bobgai can be explained by the fact that in L the word is divided by the end of a line, ‘Bol-gai’, which makes it possible that the hyphen next the l has been taken together as forming a b. In the case of Crumgring the suspension-stroke in L has been interpreted as an m: the n in L’s Crungring is, of course, only my interpretation of it. In the case of Lugtiag, the offending i appears where the scribe has experienced some difficulty, for there is evidence of an erasure above and below it. As for Dare Dromor, the word-division here is unstable in other manuscripts. The only truly erratic readings are the addition of e in line 28 and the addition of a in line 79.

Summary of main patterns of variants

Attrition like this cannot account for all the variants in the other manuscripts. There are twenty-two occasions where A and C or C and F share variant readings against L. Many of these cannot readily be explained as unintentional misreadings: they frequently involve the presence or absence of two letters in particular, h and g. Before aspersions are cast on the competence of the scribe of L, it should also be noted that an unusual level of variation involving the presence or absence of h and (to a much lesser extent) g is apparent in C and also in the lost exemplar shared by A and F (which we can call a).[25] The presence or absence of h in combination with c or t accounts for eight of the ten areas of ambiguity indicated by braces in the edition (below). This cannot be explained as the result of simple copying mistakes. There must be a different phenomenon at work: intentional variation between particular pairs of consonants.

On inspection this is largely limited to the following:

(i) c and ch,

(ii) t and th, and

(iii) More rarely, g and ch.

It appears that the scribes of L, C and a (the lost exemplar shared by A and F) in particular have allowed themselves some flexibility about which consonant within each pair they should opt for. What this means is that, when the scribes of L, C and a saw th in their exemplar, they felt free on occasion to use t, and vice versa (which would explain Scoth where Scot might be expected); or, when they saw g, they might opt to change this to ch, and vice versa; or they might equally on occasion change ch to c, and vice versa (which would explain ch in Erch and Sencormach). Similar variation can be detected in T. All in all this ‘intentional variation’ is limited to dental and velar fricatives (as in ‘th’ in modern English, ‘ch’ in modern Scots, and ‘gh’, like ‘ch’ but further back in the mouth). It may also have included unvoiced dental and alveolar plosives in final position (‘t’ and ‘c’).

This phenomenon of a scribe varying the spelling of these particular phonemes is most vividly illustrated by the copying of one word twice (in error) in T, once as Romaich, the other time as Romaig. In his exemplar the word must have appeared only once with either ch or g (representing a velar fricative): the scribe has therefore felt at liberty on one occasion to change ch to g or vice versa. In sum, it seems that scribes could give themselves some freedom of choice in which one of these pairs to apply; other scribes (like the one responsible for R) preferred not to. In the case of these ‘intentional’ variants, as it were, a further level of instability would be added by the possibility of inadvertent changes, particularly when t was misread as c, or vice versa.

Edition of the genealogy of the king of Scots in Ralph of Diss’s Imagines Historiarum

An edition of the genealogy that is sensitive to the ambiguities posed by the manuscript witnesses needs to find a way of taking account of these ‘intentional’ variants. A stemmatic approach which aims to reconstruct a lost exemplar would hardly seem to be feasible or appropriate. This is not to say that a stemma would be of no use at all. If it was based on privileging occasional readings that were clearly closer to the original Gaelic, it could serve to raise doubts about assuming that L is, in fact, the archetype of all these manuscripts. For example, in line 17, the ‘o’ shared by A’s ‘Douenald’ and C’s ‘Dounald’ against the ‘u’ in L’s ‘Duuenald’ is closer to medieval Gaelic Domnall, which suggests that in this case A and C appear to be superior to L. Another instance is line 40, where the additional ‘m’ shared by C’s ‘Drommor’ and A’s ‘–drommor’ (in F: ‘Drommor’) against L’s ‘-dromor’ is closer to Gaelic Dornmóir: it may be suggested that ‘Dornmoir’ become ‘Drommor’ in the exemplar due to metathesis.

This edition gives variants for L, A, C, F and T, but not R. There is no base manuscript: the reading given in the text is shared (if possible) by at least two of L, C or A/F/T agree. Braces { } have been used to indicate areas where there are no shared readings. No attention is given to areas where L stands alone against other witnesses unless (in one instance: line39) there is a text-historical reason for preferring L’s reading. Consonants potentially subject to intentional variation (representing fricatives and final plosives, as explained above) have been highlighted in bold, regardless of whether any variants occur. The only exceptions are (i) in established Latin forms (Constantini, Dunecani, Malcolmi); (ii) initial c and g (including eponyms where these have been recognised as a separate word), which appears to be regular,[26] unlike initial t/th: there are instances of th where t would be expected; also, g has not been indicated where it is in combination with n or where it is used consistently in an orthodox way in a name which appears more than once: Fergus, Engus/Enegusa/Enegussa.

For ease of reading, a hierarchy of citations of witnesses of a is deployed:

• If F and T are not cited this is because at least one or other agrees with A;

• After line 91 (when A ceases) T is not cited where it agrees with F (or is illegible), or where F agrees with both L and C.

The following will also not be noted:

• Variant readings which have no phonological significance, such as j for i or v for u (and including the erasure of f[ilius] in line 37 in L);

• Obvious copying errors, such as the misreading of minims, the confusion in lines 36 and 37 in A, F and T, and the omission of r in Forgso in A and F (it is illegible in T);

• If any of L, C or A (or, after line 91, F) separates name from epithet, that will be preferred without noting which MSS merge name and epithet and which do not.

The Genealogy of the King of Scots from William the Lion (1165–1214) to Noah