Robert de Torigny and the family of Gunnor, Duchess of Normandy
by
Todd A. Farmerie

[Modified from an article which appeared in Dec 1996 on soc.genealogy.medieval]

There have been many requests for information on the various Norman relationships compiled by Robert de Torigny. This is an attempt to summarize and harmonize several recent works on some of the lines:

Elisabeth M C van Houts. Robert of Torigni as Genealogist. in Studies in Medieval History presented to R. Allen Brown, p.215-33.

Kathleen Thompson. The Norman Aristocracy before 1066: the Example of the Montgomerys. in Historical Research 60:251-63.

K S B Keats-Rohan. Aspects of Torigny's Genealogy Revisited. in Nottingham Medieval Studies 37:21-7.
Robert de Torigny, writing after the Norman Conquest, recorded the genealogical traditions which tied many of the Norman nobility to the family of Gunnor, first mistress of Richard I, then Duchess of Normandy. He reported the tradition that Richard had become infatuated with the wife of one of his foresters, but being the pious wife, she substituted her sister Gunnor, much to everyone's satisfaction. He proceded to name the siblings of Gunnor, and also indicated the she had numerous nieces, who are left unnamed, but whose marriages and descendants are provided.

The genealogical information contained in his account has at various times been praised and condemned, but recent opinion seems to favor the view that, while minor errors abound, the genealogies accurately represent a tradition of shared descent that may account for the rapid rise of these nobles.

The parentage of Gunnor and her siblings is unknown. While some sources call her father Herfastus, this was in fact the name of her brother. She has also been claimed as daughter of the Danish royal family, but there is no evidence for this, and the context of her coming to the attention of Richard I and the family's subsequent rise to power militates against her being a royal daughter. Douglas argued (in a 1944 English Historical Review article on the family of William Fitz Osbern), based on the donations of brother Arfast to the monastery of St. Pere, that the root of the family was in the Cotetin region of Normandy, but van Houts has suggested that the Cotetin land was granted to Arfast, rather than inherited by him. Thus we are left with the more ambiguous statements of Torigny and others that she was a member of a Norman family of Danish origins.

The only known brother of Gunnor was Arfast/Herfast, of whom we gain what little insight we have from a trial of heretics conducted by King Robert II of France. Arfast testified that he had pretended to join the sect, all the better to denounce them when the time arose. He later donated lands to the monastery of St. Pere, to which he retired. He had at least two sons: Osbern, who was steward to the later Dukes, and was murdered by William de Montgomery while defending the young Duke William; and Ranulf, known from charters. Osbern maried a niece of Richard I (the daughter of his half-brother) and by her was the father of the Conquest baron William Fitz Osbern.

Gunnor had at least three sisters, of which the oldest appears to have been Senfria (Seinfreda), who was wife of the (unnamed) forester from the area of St. Vaast d'Equiqueville, and it was her charms which are said first to have attracted the attentions Duke Richard I. She appears to have had at least one daughter, Joscelina, wife of Hugh de Montgomery. (Torigny makes Joscelina daughter of another sister, Wevia, but a contemporary of Torigny, in demonstrating the genealogical impediment to a marriage of a bastard of Henry I to a Montgomery descendant specifically calls Joscelina's mother Senfria, and the inheritance by the Montgomerys of large holdings suggests that Joscelina was a significant coheiress to her parents, which does not match Wevia's family where the two sons would be expected to acquire most of the family land.) Hugh de Montgomery and Joscelina had a son Roger, but contrary to Torigny's statements, he was not the Conquest baron of that name, but instead his father. By a wife possibly named Emma, Roger had: Hugh; Roger (who married Mabel of Belleme and played a significant role in pre-Conquest Normandy); William (who murdered cousin Osbern); Robert, and Gilbert.

Duvelina, a second sister of Gunnor, married Turulf de Pont Audemer, son of a Norman founder Torf, and uncle of the first of the Harcourts. They had at least one son, Humphrey de Vielles, who in turn was father of Roger de Beaumont, another Conquest-era baron.

Wevia, the only other sister of Gunnor named by Torigny, married Osbern de Bolbec (who is otherwise unknown to history). They had at least two sons: Walter Giffard, ancestor of the English Giffard/Gifford families, and also, through a daughter, of the Clare family; and Godfrey, whose son William de Arques had two daughters and co-heiresses.

Torigny indicates that Gunnor had numerous nieces, naming the descendants of several of them, but usually not naming the nieces themselves or their parents. As has already been seen with niece Joscelina, the accounts of these families are more difficult to harmonize with other available sources.

One niece is said to have married Nicholas de Bracqueville, and to have had William Martel and Walter de St. Martin. As to Martel, there seems to have been a connection to Bracqueville, since Hawise, daughter of Nicholas married Hugh de Wareham, son of a Grippo. Hugh had a brother Geoffrey Martel, but beyond this no recent analysis provides any insight as to the descent of the later Martels. Walter de St. Martin is even more of a problem, since elsewhere Torigny incorrectly makes him brother of William de Warenne, but the ancestry given there is clearly false. Thus it is not clear that Torigny knew the exact connection of Walter, and there is no evidence to help clarify his true origins.

A second niece is said to have married Richard, vicomte of Rouen (who was son of Tesselin). He had a son Lambert of St. Saens, whose son Helias married a bastard daughter of Robert II of Normandy. (If the connection here given is correct, then these two were within the prohibited degree, which may throw doubt on the relationship, or simply suggest that the relationship did not come to light at the time.) Based on later interactions between Montgomery and Warenne (thought to be related to this branch) it has been speculated that this niece was sister of Joscelina, which is possible but unsupported.

It appears to be through this family that the relationship of two more Norman barons come into play, but not exactly as Torigny presents it. He shows yet another niece marrying Ranulph de Warenne, and by him having William de Warenne and Roger de Mortimer. This is clearly untrue, because Roger appears to have been a generation older than William. The solution appears to be that Torigny (as he had done with the Montgomerys) compressed two people, a father and son of the same name, into one individual. Ranulph de Warenne (I) appears to have married Beatrice, sister of Richard, vicomte of Rouen, and thus sister-in-law of one of Gunnor's nieces (thus it would appear that this family actually does not descend from a relative of Gunnor's, but is genealogically linked to some of her descendants) and had sons: Roger (de Mortimer) and Ranulph de Warenne (II), who in turn was father of another Ranulf (III) and of William de Warenne.

Finally, Torigny states that a niece married Osmund de Centumvillis, vicomte of Vernon, and had a son Fulk de Alnou, and a daughter whose son was Baldwin de Reviers. Much debate has focussed on the attempt to identify these men, but in the latter case, clearly a connection to the Reviers/Vernon Earls of Devon is intended. The precise nature of the relationship is more difficult to pin down. It would seem that the first Earl Richard de Reviers and his brother Hugh were sons of a Baldwin, who had brothers Richard de Vernon (app. d.s.p.) and William Fitz Hugh de Vernon. (William, who was perhaps a uterine half-brother, had by wife Emma a son Hugh, often confused with the brother of Earl Richard. It is this error that has led to the statement that Emma was the relative of Gunnor, which derives from a set of relationships hypothesized in Complete Peerage (CP, under Devon) and predicated on her being mother of Hugh, brother of Earl Richard, an untrue relationship, and on Richard being nephew of William Fitz Osbern, which is discussed below.) If Baldwin, father of Earl Richard, was the same as the grandson of Osmund de Centumvillis this would complete the picture, but one more relationship invites comment. Earl Richard is said by an early source, cited by CP, to be nephew of William Fitz Osbern. If the stated connection with vicomte Osmund is correct, then Baldwin de Reviers would have been too closely related to William Fitz Osbern to have married his sister. (An alternative solution, that the wife of vicomte Osmund was sister of William Fitz Osbern, and hence grandniece of Gunnor, is chronologically impossible.) I suspect that this tradition records the memory that William Fitz Osbern was an older male relative of Richard, rather than a precise genealogical relationship.

The work of Robert de Torigny thus provides a valuable source for the genealogical origins of the immediate pre-Conquest Norman aristocracy. When it has been possible to compare the information with other sources, some inconsistancies are found, but it is unclear whether these represent errors of Robert, or inaccuracies in the genealogical traditions he was recording. In most cases, an in-depth study of the available material has enabled modern historians to satisfactorilly reconstruct the descents from Gunnor's family and provide a representation of the true relationships among these early Norman families.

Todd A. Farmerie