The earliest Welsh genealogies: textual layering and the phenomenon of ‘pedigree growth’

This article examines the ways in which early medieval genealogical texts might be augmented over time in order to reflect changing political situations. Two early ninth‐century tracts from the kingdoms of Powys and Dyfed in Wales are taken as case studies. Textual and chronological problems with the tracts are discussed, and contexts are proposed for the circumstances of their composition. It is suggested that each of these tracts stands at the head of a process of ‘pedigree growth’, whereby, during the course of textual transmission, the genealogical content of each tract was extended both backwards and forwards in time.

Gwynedd (d. 1282), for example, as found in the fifteenth-century manuscript London, British Library, Harley 673, includes no fewer than ninety-four generations, inclusive of both Llywelyn and God. 8 The same occurred elsewhere; Ailred of Rievaulx, drawing on a pedigree of the West Saxon kings that had been in writing since at least the ninth century, was able to trace the ancestry of Henry II back fifty-six generations to Adam. 9 If genealogies were disposed to grow over time, it follows that earlier genealogical texts should generally be shorter in length. This is indeed exactly what we find with some of the earliest extant Welsh genealogical texts, embedded within the Harleian genealogical collection. The latter collection originated as an illustrative appendix to the ninth-century Cambro-Latin composition Historia Brittonum, and was probably composed sometime in the third quarter of the ninth century, possibly around 858. 10 By the time the collection reached its extant form, in c.954, it had been interpolated with a number of smaller genealogical texts taken from other sources. 11 Since the latest stage of this interpolation happened in the church of St Davids, in south-west Wales, the interpolated form of the genealogical collection may be called the 'St Davids genealogical collection'. Two of the interpolated texts, which form the subjects of the present paper, concern the kingdoms of Powys and Dyfed in the early ninth century. I call the two texts the 'Powys tract' and the 'Dyfed tract'. These are two of the earliest genealogical texts that have survived from early medieval Wales, and they are probably indicative of the form once taken by the earliest stratum of Welsh genealogical records.
The two texts share a number of traits. For a start, both express relationships between individuals using Latin rather than Welsh terminology (e.g. Latin filius for Old Welsh map, 'son'). In this sense they sit oddly in the extant Harleian collection, much of which uses Welsh for this purpose. 12 They do not conform to the standard pedigree model. Neither text extends for more than five generations; instead, more information is given about collateral relatives than is usual in Welsh genealogical texts written before the thirteenth century. A longer pedigree of the expected sort probably did form part of the Dyfed tract originally, as we shall see, but this would not have extended back as many generations in the early ninth century as some of its derivative versions in later Welsh genealogical texts eventually came to do. In both cases, however, it is clear that the longer pedigrees that later came to be associated with the kingdoms of Powys and Dyfed were textual developments that grew out of more limited genealogical records of the kind represented by the two tracts. As the text stands, two lines of descent are given, one of three and one of four generations, bound together by a common ancestor, Elized (modern Welsh 'Elise'), in the fourth generation. The link with the kingdom of Powys is not made explicit in the text, but may be adduced through a comparison with the well-known Pillar of Eliseg. This monument was erected near present-day Llangollen in north-east Wales, probably in The inscription then goes on to explain that King Cyngen had ordered the monument to be erected for his great-grandfather Elise, who 'necxit hereditatem Pouos' ('united the inheritance of Powys'). Here, as in the Powys tract, a certain Elise is cast in the role of dynastic progenitor. Although Elise and his son Brochfael are the only figures shared by the Powys tract and the inscription, Elise's consistent role in the two texts helps to substantiate the connection between the lineages of the Powys tract and the kingdom of Powys.
Later in the inscription, the remoter ancestral figures Brydw, Vortigern, and Severa are mentioned, who provide links to the sources of legitimacy represented by Magnus Maximus and St Germanus. 17 Although some sections of the inscription were no longer legible to the seventeenth-century antiquarians whose transcripts we rely upon for our text, it is not necessary to assume that the inscription ever included a specific agnatic line of descent from Gwylog back to Brydw, especially considering the lack of space on the monument. 18 One may note Molly Miller's observation that the Britto/Brutus of the Historia Brittonum did not need to be genealogically linked to any specific descendants in order to fulfil his aetiological function. 19 Perhaps the Pillar of Eliseg's Brydw (Britu) once similarly functioned as an eponym for the Britons of Powys, affiliated in an unspecific way to Cyngen's dynasty. 20 The form of the name Grippi in the Powys tract suggests that some kind of textual error has been made at this point. The form should either be Grip(p)ri (modern Welsh 'Griffri'), perhaps with an Insular long r having been misread as a p, or Grippiud (modern Welsh 'Gruffudd'), with two letters having been omitted from the end of the word. 21 Egerton Phillimore suggested that this Grippi should be identified with a certain Gruffudd who appears in the Welsh Latin 'Harleian chronicle', in the annal for 815: 22 Gripiud filius Cincen dolosa dispensatione a fratre suo Elized post interuallum duorum mensium interficitur.
After a space of two months Gruffudd son of Cyngen is killed by his brother Elise through deceitful planning.
It is possible that the Elise son of Cyngen in this annal should be equated with the Elise son of Cyngen in the second line of the Powys tract. The names Elise and Cyngen are, in this period, characteristic of Powysian dynasts. 23 In view of the textual error implied by the form Grippi in the first line of the Powys tract, it is also possible, as Phillimore suggested, that Grippi can be identified with the annal's Gripiud, implying that the Powys tract has omitted the names of Grippi's father Cyngen and grandfather Brochfael, in addition to the end of Grippi's name. This identification is speculative but plausible, and would explain the absence of a brother of Elise called Gruffudd in the second line, since he had already been mentioned in the first line. Phillimore, however, went one stage further, suggesting that Cyngen, the father of Elise and Gruffudd in the annal, was the Cyngen, king of Powys, who erected the Pillar of Eliseg. 24 Such an identification requires the insertion of filii Catel after Cincen into the second line of the Powys tract, as well as the insertion of a longer passage, -ud filii Cincen filii Catel filii Brocmail, following the problematic Grippi in the first line. Phillimore's emendations of the text, which have been accepted by many subsequent scholars, 25 produce the scheme in Fig. 1 (note that in Figs 1-3, names in HG 30 are underlined, names in HG 31 are in bold and names on the Pillar of Eliseg are italicized).
There are a number of problems with Phillimore's reconstruction, not the least of which is chronology. 26 Cyngen, king of Powys, died in 854, forty-six years after the death of his father Cadell in 808, and yet, according to this scheme, he had two politically active adult sons already in 815. 27 Instead, I suggest that one of the two following solutions presents a sounder result:   (Fig. 2). ii) Phillimore was incorrect to identify Grippi with the Gruffudd ap Cyngen of the annal for 815, meaning that no substantial emendation of the Powys tract is required, as Patrick Sims-Williams has suggested (Fig. 3). 28 Phillimore's identification of the [E]lized filius Cincen of the Powys tract with the man of the same name in the annal for 815 may nevertheless be correct, and Gruffudd may have been omitted from the list of Elise ap Cyngen's brothers in HG 31 either because Elise had already killed him 29 or for some other reason, such as the two having been born from different mothers. The important point in the present context is that neither the Cyngen of the Pillar of Eliseg (d. 854) nor his father Cadell (d. 808; possible brother of the Cincen of HG 31) appear in either reading. This may be because Cyngen ap Cadell's pedigree is found elsewhere in the Harleian collection, but it may equally be because there was a period (represented by the 815 annal) in the generation preceding Cyngen's kingship of Powys in which his branch of the family was not paramount, and others instead warranted inclusion in the genealogical record. This illustrates the limits of our knowledge of Welsh dynastic politics in the early Middle Ages, and emphasizes the dangers of reading long, linear pedigrees as simple king lists.
The Powys tract was presumably written in a monastic house in Powys favourable to Elise's dynasty. Since the Pillar of Eliseg advertises a connection between this dynasty and St Germanus, known in Welsh as St Garmon, it may be suggested that the monastic house responsible for the Powys tract was located in Llanarmon-yn-Iâl ('the church enclosure of St Garmon in Iâl'), near the Pillar of Eliseg. It is notable in this respect that a certain Llanarmon, most probably Llanarmon-yn-Iâl, is singled out as one of the praecipuae ecclesiae of north Wales to which Gruffudd ap Cynan, a later ruler of Gwynedd, donated money on his deathbed in 1137, as reported in his twelfth-century vita. 30 The lack of any long ancestry for the dynastic founder Elise in the Powys tract may be indicative of the text's early date (compare the lack of any explicit long pedigree for Elise on the Pillar of Eliseg). Just such an ancestry was later given to Cyngen, Elise's great-grandson, as may be found in §27 of the Harleian genealogies, where his line is traced back through Elise to a certain Cadell. It seems probable that this pedigree was included in the source of the Harleian genealogies in order to illustrate the claim made in the Historia Brittonum that Powys was governed by the descendants of Cadell Ddyrnllug 'usque in hodiernum diem' ('to this day'). 31 As has been noticed and discussed many times before, the Historia Brittonum's account of the dynastic origins of the kingdom of Powys seems to be at variance with the account on the Pillar of Eliseg; the Historia Brittonum states that the kings of Powys were all descended from Cadell, who was blessed by Germanus, whereas the extant text of the Pillar of Eliseg makes no reference to Cadell but mentions Vortigern, whose son Brydw was, again, blessed by Germanus. 32 Patrick Sims-Williams has argued that the parts of the Pillar's inscription that were illegible to the seventeenth-century transcribers may have introduced Vortigern in some way that did not conflict with the account in the Historia Brittonum; 33 however, it seems difficult to imagine that the Pillar would have claimed that Vortigern's son Brydw had been blessed by St Germanus and was the grandson of Maximus had there been no implication that Cyngen's dynasty was connected with Brydw, regardless of the question of whether or not the inscription ever included a specific pedigree demonstrating this. 34 I would agree with those who have argued that the contrast between the two texts was somehow a consequence of the dynasty of Elise ap Gwylog having wrested some power in Powys away from the Cadelling dynasty. 35 The Historia Brittonum's hagiographical account about St Germanus establishing Cadell and his heirs as the eternal rulers of all Powys would, according to this view, be at least partially anachronistic by the first half of the ninth century, when the Pillar of Eliseg was erected. The Historia Brittonum's source for its Germanus material was apparently a saint's Life, named at one point as the Liber beati Germani; perhaps this lost Life was composed before the rise of the dynasty of Elise ap Gwylog. 36 The difference between the Pillar of Eliseg and the Historia Brittonum might also be a product of the way in which the Liber beati Germani was composed. As has been said, the location of the Pillar of Eliseg suggests that its reference to St Germanus was prompted by the proximity of the church of Llanarmon-yn-Iâl. Traditions from the same church were almost certainly responsible for the Cadell story in the Historia Brittonum, via the Liber beati Germani. 37 However, it is possible that the Liber beati Germani was also the source for a further genealogy embedded within the Historia Brittonum, one which appears immediately after the account of the interaction between Germanus and Vortigern. This genealogy is traced back to Vortigern, but it does not begin with a member of the dynasty of Elise ap Gwylog; rather, it begins with a certain Ffernfael, 'qui regit modo in regionibus duabus Buelt et Guorthigirniaun' ('who rules now in the two regions of Buellt and Gwerthrynion'). 38 The text locates the descendants of Vortigern in mid-Wales, rather than near Llanarmon-yn-Iâl in the north east, as the Pillar of Eliseg implies. This account may derive from the church of St Harmon ('Llanarmon' in Welsh), another important church of St Germanus, but one located within the bounds of Gwerthrynion. 39 This is supported by R.W.D. Fenn's argument that Germanus's forty-day stay in Gwerthrynion during his pursuit of Vortigern in the Historia Brittonum was intended to allude to the foundation of the church of St Harmon. 40 The Liber beati Germani thus incorporated traditions from more than one church of Germanus, meaning that it is not at all certain whether it was composed from the specific perspective of the church of Llanarmon-yn-Iâl, as probably was the nearby Pillar of Eliseg. Perhaps this accounts for the differing portrayals of the origins of the kings of Powys in the Historia Brittonum and the Pillar of Eliseg.
The long pedigree tracing Cyngen back to Cadell Ddyrnllug had not necessarily been created by the time that Cyngen's father, Cadell, received his name; 41 David Thornton has noted the process whereby intrusive dynasties appropriated the names of their vanquished dynastic rivals, without necessarily claiming direct agnatic descent from their rivals' dynastic ancestors. 42 The long pedigree found in HG 27 probably reached its extant form during its transmission as part of a genealogical appendix that sought to explain and clarify the Historia Brittonum. Redactors of the appendix had to reconcile the statement that Powys was ruled by the descendants of Cadell with the reality of the contemporary dynasty of Elise ap Gwylog. A pedigree of Cyngen identical to that on the Pillar of Eliseg seems to have been available as a basis, and at some point this was extended backwards using a Cadelling pedigree similar to that of the Cadelling king Selyf (d. c.615) 43 found among the interpolated north-eastern material in the Harleian genealogies ( §22). 44   The point at which Cyngen's own pedigree seems to have been joined to the Cadelling pedigree is marked by the biblical name Eli, which is known to have been used among the Britons elsewhere. 46 Later versions of the pedigree name the same figure as Beli, though this might be because 'ap Eli' was, in Welsh, homophonous with 'ap Beli'. 47 In the present context, the consecutive alliterating names Eli and Eliud might be a pair of filler names, inserted to join together two originally separate pedigrees, since in HG 22 Selyf occupies the same genealogical position as Eliud. The same phenomenon may be observed in the invented sections of many other Welsh pedigrees; compare, for instance, the pairs of names (such as Nyfed and Dyfed) in the Dyfed pedigree discussed below. 48 As has been observed before, other genealogists would try to reconcile this artificial Cadelling pedigree with Cyngen's own proclaimed connection with Vortigern, creating, as a consequence, ever longer pedigrees. 49 This is a clear example of the process of 'pedigree growth' outlined at the beginning of this paper. At the head of the process stands a text like that preserved as the Powys tract in the Harleian genealogies, which records, no doubt accurately, a genealogy extending over no more than four or five generations. Over time, successive attempts to alter and expand the dynasty's genealogical record led to the growth of the 'Powys pedigree', which began to assume a textual existence of its own. One can witness the same process in more detail in the genealogical records of the kingdom of Dyfed, in south-west Wales, to which we now turn.

The Dyfed tract
The Dyfed tract provides a comparable example of an early ninth-century text standing at the head of a long process of pedigree growth. The text survives as sections 13-15 of the Harleian genealogies, as follows:  JC 8 includes the patrilineal ancestry of Nowy, whose father, we are explicitly told, was one Tewdwr, son of Rein. It is likely that some, though not all, of JC 8's patrilineal pedigree appeared in the common source of HG 15 and JC 8.
JC 8 provides a good example of pedigree growth, both in the chronologically more recent and in the chronologically remoter parts of its text. The beginning of the text has been extended forwards by four generations, from Gruffudd ap Nowy, one of the three brothers in HG 15, to Tewdwr ap Griffri, a member of a family that ruled in Brycheiniog, in south-central Wales, in the mid-tenth century. 54 There was also a need to alter the earlier generations of the dynasty due to Tewdwr's family having become rulers of Brycheiniog rather than Dyfed. For this reason, Tewdwr's distant ancestor, Caden, is made to be the son of one Ceinddrech, daughter of Rhiwallon, whose own patriline is conveniently traced to Brychan, the eponymous founder of Brycheiniog. The link was almost certainly contrived.
Nevertheless, the section of JC 8's pedigree in which the male ancestors of Nowy are traced back four generations further than in HG 15 may well have appeared in the ultimate source in the early ninth century. The inclusion of at least Nowy's father Tewdwr and grandfather Rhain would have made the connection between HG 15 and HG 13-14 explicit, just as the connection between the two parts of the Powys tract is made explicit through the inclusion of the common ancestor Elise. Remarkably, we can be certain that just such a pedigree, as partially found in JC 8, once existed independently. A pedigree traced back from a certain Tualodor, who is to be identified with either Tewdwr or his brother Tewdos, is found in the Old Irish tale Indarba na nDéisi, 'The Expulsion of the Déisi', which has consequently been dated to the brothers' lifetime, in the middle of the eighth century. 55 The pedigree is found only in the 'earlier' version of the tale, extant in four manuscripts. 56 Some of the name forms in this pedigree imply that it was taken from an Old Welsh source. 57 Since the extant form of JC 8 suggests that the source that it shares with HG 13-15 included at least the first five generations of this pedigree, it is not at all unlikely that their common source included a longer pedigree similar to that found in Indarba na nDéisi. The question is why no such pedigree exists in HG 13-15. The answer seems to be that the pedigree had already been included elsewhere in the St Davids genealogical collection. At the same time that HG 13-15 was interpolated into the St Davids collection in the mid-tenth century, a pedigree was inserted into second position in the collection (now HG 2) which traced the Demetian ancestry of Helen, mother of Owain ap Hywel Dda, the king ruling south-west Wales (950-88) at the time of the interpolation. 58 The Demetian pedigree of HG 2 runs largely parallel to the pedigree in Indarba na nDéisi. In another example of pedigree growth, the differences are found at the beginning and the end. In the first part of HG 2, Helen's ancestry is traced back to Tewdos ap Rhain, who appears in the Dyfed tract in HG 13. In the latter reaches of the pedigree, other changes have been made that are best discussed only once all the extant versions of the pedigree have been laid out in parallel, as is done in Table 1. The table has been arranged so as to show the correspondences across the columns; a gap in a particular row does not indicate that there is a gap in that particular text, only that there is no name in that position corresponding to the other names in that row.

The Dyfed pedigree
The four versions of the Dyfed pedigree have been discussed in relation to one another before, not least because of the great variation that they display in their earlier reaches. 59 What is less often asked is why they should be almost perfectly consistent as far back as Tryffin (fl. s. v/vi?). The eleven generations back from the brothers Tewdwr and Tewdos to Tryffin appear almost without variation; the only differences of note are the (probably accidental) omission of Rhain and the insertion of Erbin, found in both the thirteenth-century Llywelyn ab Iorwerth genealogies and in Jesus 20 genealogies, § §12-13. By contrast, beyond Tryffin there are dramatic differences in the selection and arrangement of the generations in each version. Some independent check seems to have prevented the generations between Tewdwr/Tewdos and Tryffin from being drastically changed or manipulated.
A cause for the unusual consistency may have been the one-time existence of an archive of early medieval documents in the monastery of St Davids. It has been suggested previously that a comparable archive was used to construct the pedigrees of the kings of south-east Wales in the Harleian genealogies. 60 Although no charters from pre-Norman St Davids are now extant, there is evidence from more than one authority for the survival of such documents into the sixteenth century. 61 Among the extracts supposedly taken from these documents in the sixteenth century are transcribed letter forms resembling the script used for the ninth-century Old Welsh charters preserved in the Lichfield Gospels;  spellings that probably pre-date the twelfth century; 63 and, as William Salesbury emphasized, a preponderance of biblical names, characteristic of the Britons in the early Middle Ages. 64 Most importantly for present purposes, the extracts include the names of some of the early kings of Dyfed who appear in the Dyfed pedigree, including 'Noe Rex Demetiae' ('Nowy, king of Demetia/Dyfed') and Nowy's father, 'Arturius Petri filius sed recentior veteri Arturio' ('Arthur son of Pedr, but more recent than the old Arthur'). 65 The position of Arthur, Nowy, and Pedr in the pedigree suggests that they lived during the seventh century. 66 There is therefore good evidence that charters were indeed written and preserved in St Davids during the early medieval period, and that these charters included in them the names of the early kings of Dyfed, occasionally alongside their patronymics. These documents may have been eventually gathered together into the lost volume referred to by John Leland as De Dotatione Ecclesiae S. Davidis. 67 Such a series of charters might have been used in the eighth century to help construct the pedigree of the kings of Dyfed back to Tryffin, and the continued existence of the archive might have prevented this section of the pedigree from changing too obviously.
There is a further indication that the evidence of the charters might have stretched back as far as Tryffin. This is provided by Gerald of Wales's list of the bishops of St Davids, which extends from St David, who supposedly died in 601, to Peter de Leia, who occupied the see (1176-98) at the time when Gerald was writing. 68 Although some names in the list, such as that of St Samson, have been inserted retrospectively, there is no reason to doubt entirely the names of the bishops between David in the late sixth century and Sadyrnfyw in the early ninth. The list, like the pedigree, is exactly the kind of document that could have been compiled using the evidence of a sequence of charters. If such a sequence of charters commenced within a generation or so after the time of St David, then Tryffin is a likely candidate for the representative of the limit of historical memory at the time that the archive was initiated. That the monastery of St Davids had an active scriptorium at this early date might be suggested by the survival of a group of four related penitential texts of possible sixth-century origin, extant in two Breton manuscripts of the ninth and tenth centuries. One of these texts is entitled excerpta quedam de libro dauidis. 69 If the stable section of the Dyfed pedigree belongs to the world of history and authentic sources, the unstable sections beyond Tryffin belong to the world of pseudo-history and retrospective invention. In the case of the Dyfed pedigree, such activities can be assigned, with some confidence, approximate dates, provenances, and causes. Figure 5 presents a stemma of the proposed relationships between three of the four versions of the pedigree laid out in the table above. The common archetype [α] of all four versions of the pedigree must have existed by the floruits of Tewdwr and Tewdos, sons of Rhain, in the middle of the eighth century, at which time a version of the pedigree was probably taken to Ireland and incorporated into Indarba na nDéisi. The latter provides the earliest witness to the archetype. It is unlikely that Tryffin's Irish ancestry, as found in the Irish tale's pedigree, was invented in Ireland, because elements of it are present in the three Welsh versions of the pedigree. At a later stage [γ], in the tenth century, a derivative of α was incorporated into the St Davids genealogical collection as the pedigree of Owain ap Hywel Dda's mother Helen. It is likely that the pedigree was transmitted to the St Davids collection via the early ninth-century Dyfed tract [β], as partially witnessed by JC 8.
Stage γ is best represented by the Harleian genealogies and the Llywelyn ab Iorwerth genealogies. The Llywelyn ab Iorwerth genealogies comprise a Welsh genealogical collection composed in the early thirteenth century and extant in manuscripts from the fifteenth century onwards. 70 One section of this text drew material from the mid-tenth-century St Davids genealogical collection. 71 Both the Harleian genealogies and the Llywelyn ab Iorwerth genealogies include the pedigree of Owain ap Hywel Dda traced through his mother Helen. It is clear from Table 1, however, that the earlier part of the pedigree must have been altered significantly in at least one of these two versions at some point between the mid-tenth century, when the St Davids collection was first created, and the texts extant in the manuscripts. Despite the general superiority of the Harleian genealogies as a witness to the St Davids collection, it would appear that, in this case, it is the Harleian genealogies that give the altered version.
There are two main points of difference between Tryffin's ancestry in the Llywelyn ab Iorwerth genealogies and the Harleian genealogies. One is that only the Llywelyn ab Iorwerth genealogies preserve the name Ewein Vreisg, which, as the name of Tryffin's father, is seemingly equivalent to the Irish Áed Brosc. 72 Ewein seems to have been 72 Patrick Sims-Williams has suggested that the Brosc element in the pedigree was originally an independent personal name rather than an epithet: P.   ) and his wife Helen, the latter of whom, as the text goes on to relate, was responsible for the discovery of the true cross. HG 2 thus begins and ends with two women called Helen, the mothers of Owain and Constantine the Great respectively, creating a neat and undoubtedly intentional symmetry. 75 This symmetry can only have been created once Helen became the focus of the Dyfed pedigree in the tenth century. The two points combine to suggest that the Llywelyn ab Iorwerth genealogies, rather than the Harleian genealogies, preserve the form of the pedigree closest to that of stage γ, when it was first inserted into the St Davids genealogical collection along with the rest of the Dyfed tract. The Harleian genealogies, on the other hand, preserve a version that was altered after stage γ. The changes visible in the Harleian genealogies were intended to bring the earlier stages of the pedigree into line with contemporary political thought. Maximus was retained as the younger Helen's ancestor, although he was brought forward in time to be the father of the eponymous Dimet, 'Dyfed'. Constantius and Helen were added in the place of Maximus as the ultimate ancestor figures. At the same time, the most obvious Irish elements in the pedigree were removed. This was probably because, by the tenth century, Dyfed was ruled by the powerful Merfynion dynasty, which claimed descent from Cunedda Wledig. 76 The latter was the man who, according to the Historia Brittonum (which accompanied the St Davids genealogical collection at stage γ), had expelled the Irish from Britain, apparent. At stage α, possibly in the eighth century, a twelve-generation pedigree from Tewdwr or Tewdos to Tryffin was constructed, perhaps using evidence from the archive of St Davids. It may have been at the same time that extra generations were added beyond Tryffin, confirming the alleged Déisi origin of the dynasty. At some point between α and γ, possibly when the pedigree was incorporated into the Dyfed tract at stage β in the early decades of the ninth century, the pedigree was extended back further again to include such figures as Clydwyn, supposed conqueror of south Wales, Dimet, eponym of Dyfed, and ultimately that great symbol of post-Roman legitimacy, Maxen Wledig. 83 Later, at St Davids in the mid-tenth century, genealogists working under the new Merfynion regime removed the visible Irish elements from the Dyfed pedigree and instead honoured their king's mother Helen by deriving her lineage from Helen, finder of the true cross and mother of Constantine the Great, perhaps implying at the same time that the latter-day Helen's son Owain should be considered a second Constantine.
Processes of 'pedigree growth' such as those explored above probably lie behind many early medieval genealogies, but rarely is one able to observe the process quite as clearly as in the case of the Dyfed pedigree. A comparative textual study such as the one presently undertaken allows the historical credibility of the long and often intractable lists of personal names appearing in genealogical manuscripts to be thrown into sharper relief. The vexed question of the historical reliability of genealogical texts as direct evidence for the early medieval period can be seen to be a complex one indeed. Rarely is a simple answer applicable to a whole genealogy, let alone an entire genealogical collection. Some sections might represent biological fact, other sections pseudo-historical edifice, and others still subsequent re-workings of the fact and the fiction combined. Until these layers are teased out of each surviving genealogical text, the historical use of medieval pedigrees can only proceed on the most tentative basis.