FEMALE Agatha

Wife of Eadweard the Exile.

Because of the length of the discussion, the article on Agatha has been divided into two pages. The other page contains the Appendices.

Agatha married Eadweard "the Exile" sometime before his return to England in 1057. She was still living in 1067, when she accompanied her children as a refugee into Scotland [ASC(D) s.a. 1067 (p. 201); Sim. Durh., c. 155 (2: 190)].

Date of birth: Say 1015×1035?
Place of birth:
Unknown.
Given the surviving evidence, the best we can hope for is to give a range of dates in which Agatha's birth probably occurred, based on the slim evidence that exists for the ages of her children. Her son Eadgar Ćtheling was said by Orderic Vitalis to have been the same age as Robert of Normandy, son of William the Conqueror ["... ducemque sibi coćvum ..." OV x, 11 (vol. 4, p. 70)]. Robert was the eldest son (and probably the eldest child) of a marriage which occurred between 1049 and 1053, and probably in 1050 or 1051 [see the page of William I], thus born probably between 1050 and 1054, and such a birthdate would fit well for Eadgar, whom the contemporary Guillaume de Poitiers calls a "puer" in referring to events of 1066 ["Regem statuerant Edgarum Athelinum, ex Edwardi Regis nobilitate annis puerum." Guillaume de Poitiers, ii, 28 (pp. 146-7); see also ibid., ii, 35 (pp. 162-3)], making it unlikely that Eadgar was born before 1050. Agatha's daughter Margaret was married to Malcolm III of Scotland, probably in about 1070 [see the page of Malcolm III for a discussion of this date]. If we assume that Agatha and Margaret did not marry before the age of 17 and did not have a child before the age of 18, then that would place Agatha's birth in 1035 or before, with dates a year or two later possible but highly improbable, and with a birth in 1030 or before providing a more comfortable margin. In the other direction, although it seems very unlikely that Agatha was older than her husband (born 1016×7), the possibility that she was born a decade earlier cannot be strictly ruled out. Jetté places the marriage of Margaret about 1067 ("about ten years" after 1057). This is probably a few years too early [Freeman (1870-9), 4: 783-7; see the page of Malcolm III]. Because Malcolm asked for the permission of Margaret's brother Eadgar to marry her, and because "enough was known about his [Edgar's] personality to separate him from the throne of England in 1066", Jetté argues that the eldest of Margaret's children was born between 1045 and 1050 "at the latest" and thus that Agatha "cannot be born after 1030 and that she was more likely born around 1020" [Jetté (1996), 420]. However, although Agatha may have been born that early, there seems to be no reason to insist on it.

Date of death: After 1067.
Place of death: Unknown.
As noted above, Agatha was still living in 1067. Ingham places her death about 1068, saying that she is thought to have been deceased before her daughter married king Malcolm [Ingham (1998b), 240 & n. 32]. The reason for this belief would appear to be the fact that Malcolm asked Eadgar for his sister's hand [ASC(D) s.a. 1067]. The Crowland Psalter has the addition of the obituary under 18 March of a person whose name starts with "A", in the same hand as additions giving the obituaries of Eadweard and his brother Eadmund [Keynes (1985), 359-60]. This could be the date of Agatha's death, but the obituary could also be that of Eadweard's mother Ealdgyth (Aldgitha).

Father: Unknown.
Mother: Unknown.
There have been a number of mutually contradictory theories regarding the origin of Agatha, and a definitive solution is still lacking. The alternatives are discussed in detail in the Commentary section.

Spouse: Eadweard "the Exile", d. 19 April 1057.

Children:
See the page of Eadweard "the Exile" for details.

MALE Eadgar "the Atheling", living 1125, claimant to the English throne in 1066.

FEMALE St. Margaret, d. 16(?) November 1093;
m. 1070×1,
Máel Coluim mac Donnchada (Malcolm III), d. 13 November 1093, king of Scotland.

FEMALE Christina, living 1086, nun at Romsey, prob. d. 1095×1100.



Commentary

The case for Agatha's parentage, which still lacks a definitive resolution, requires a detailed discussion. First, a chronological outline of some of the main developments in the research of Agatha's origins will be given, concentrating on the more recent period. Second, a transcript of the more important primary sources will be given. Third, the numerous theories regarding Agatha's parentage will each be briefly described. This will be followed by the main discussion, and a page of appendices discussing related matter.

Researching Agatha: a chronology

The problem of Agatha's origin has been researched often, and no attempt is made here to include all modern scholarship on the subject in this mainly chronological outline. Certainly, this outline will be more complete for the more recent period, and it is quite likely that earlier modern references than the ones given here could be given. For convenience of reference, some of the main hypotheses which have been proposed have been given labels such as the "German Hypothesis" or the "Russian Hypothesis" to simplify the discussions below.

Late medieval writers were generally content to repeat the vague or incorrect accounts of one of the early medieval authors discussed below. Finally, some scholars attempted to deduce a specific parentage for Agatha from the available evidence. In 1763, György Pray argued that Agatha was a daughter of bishop Bruno of Augsburg, brother of emperor Heinrich II [Pray (1763), 1: 27-8]. I refer to this hypothesis as the "Bruno Hypothesis" in the discussions below. In 1778, Daniel Cornides concluded that Agatha was a daughter of king István (Stephen) I of Hungary by his wife Gisela, sister of the emperor Heinrich II [Cornides (1778), 232-9]. This theory will be called the "Hungarian Hypothesis". In 1779, István Katona supported the Bruno Hypothesis, as did Peter Friedrich Suhm in 1787 [Katona (1779), 1: 260-3; 2: 97-107, not seen by me, cited by Herzog (1939), 1; Suhm (1787), 3: 726]. This hypothesis was noted (but not explicitly endorsed) by Lappenberg (who called Agatha a relative of the emperor) in 1834 [Lappenberg (1834-81), 2: 243 n. 4] and by Thorpe in 1848 [John Worc. 1: 181 n. 3]. In 1877, in his History of the Norman Conquest of England, Edward A. Freeman concluded that Agatha was "most probably a niece" of emperor Heinrich II [Freeman (1870-9), 2: 376, 671-2]. In 1879, Harry Breßlau concluded that Agatha was a daughter of Stephen of Hungary, i.e., the Hungarian Hypothesis [Breßlau (1879-84), 1: 102 n. 1]. In 1938, Sándor Fest also argued that Agatha was a daughter of Stephen of Hungary [Fest (1938)]. In 1939, in a long article published in the Hungarian genealogy and heraldry journal Turul, József Herzog introduced what is here called the "German Hypothesis", which suggests that Agatha was a daughter of one of the maternal half-brothers of emperor Heinrich III [Herzog (1939)]. Unfortunately, I am unable to read the Hungarian in which this work is written, so I must depend on other reports about what Herzog said. Von Redlich and Moriarty both seem to suggest that of the three half-brothers of Heinrich III, Herzog preferred the candidacy of Ernst II, duke of Swabia, as the father of Agatha [Redlich (1940), 107; Moriarty (1952), 52]. Vajay states that Herzog "hesitates between Liudolf of Westfiresland and Ernest of Swabia, as possible fathers for Agatha." [Vajay (1962), 79, n. 34] In a short article in 1940, Marcellus von Redlich mentioned the problem of Agatha's origin, listed the candidates for Agatha's father of which he was aware (Emperor Heinrich II, Bruno of Augsburg, Stephen of Hungary, Salomon of Hungary, Iaroslav of Russia, the half-brothers of Heinrich III), and stated as his preference the version of the German Hypothesis in which Ernst is the father (called here the "alternate" version of the German Hypothesis), giving brief reasons (based mostly on secondary sources) for rejecting the others [Redlich (1940)]. In 1952, G. Andrews Moriarty discussed the problem of Agatha's origin, mentioning the alternatives that had been proposed by Freeman, Fest, and Herzog. After rejecting the others, Moriarty stated that Fest's conclusion that Agatha was a daughter of Stephen of Hungary was "highly probable and conclusive." [Moriarty (1952), 60] In 1954, R. L. Grćme Ritchie, in an appendix to his book The Normans in Scotland, considered the parentage of Agatha, and after rejecting the Hungarian Hypothesis, concluded that the Bruno Hypothesis was "perfectly tenable" [Ritchie (1954), 392]. In 1962, Szabolcs de Vajay presented arguments against the Hungarian Hypothesis (mainly citing only secondary sources written in Hungarian), and supported the variant of Herzog's theory in which Liudolf of Braunschweig was presented as the father of Agatha [Vajay (1962)]. The paper was widely cited, and a significant number of authors regarded the matter as having been settled by Vajay's paper. This is the "main" variation of the "German Hypothesis". A 1971 paper by Vajay has more information on Liudolf's supposed daughters, but contains nothing new relevant to Agatha's parentage [Vajay (1971)]. Although I have become convinced from my research that the German Hypothesis is the most likely alternative from among the numerous weak choices, Vajay's discussion and documentation is inadequate on several points, especially with regard to the family of the empress Gisela (Liudolf's mother). In 1975, Gerd Wunder, accepting Vajay's version of the German Hypothesis, suggested that Agatha had had an earlier marriage to Vladimir of Novgorod (1020-52), son of Iaroslav [Wunder (1975)]. This would explain how Agatha arrived in Russia, where Eadweard is believed to have been married, but Wunder conceded that there was no source to prove his unconvincing theory. In 1984, Gabriel Ronay published an article on Eadweard the Exile [Ronay (1984)] and then followed it up with a book which went into much more detail [Ronay (1989)]. Both the article and the book argue for the German Hypothesis. The documentation is usually inadequate, and the author frequently takes liberties with his sources. The book often reads more like historical fiction than history. In 1996, René Jetté published an article in the New England Historical and Genealogical Register suggesting that Agatha was a daughter of Iaroslav I of Russia [Jetté (1996)]. At the time the article appeared, it was widely thought that this theory was a new one, but Pavsic later cited an example to show that the theory had been around since at least the nineteenth century [Pavsic (2001), 82 n. 67]. Jetté's theory is here called the "Russian Hypothesis".

Jetté's article appears to have opened the floodgates, for in the next ten years there appeared as many articles on Agatha's origin as had appeared in the previous sixty years. It would seem that at least part of this surge in activity can be attributed to the internet. In 1995, the year before Jetté's article appeared, the internet newsgroup soc.genealogy.medieval and the internet mailing list GEN-MEDIEVAL made their first appearance. While soc.genealogy.medieval and GEN-MEDIEVAL are technically different entities, they are "gated" so that all messages from one are sent to the other, so that they act together as one big message board. Even before the appearance of Jetté's article, someone who was familiar with Jetté's research was posting "teasers" to soc.genealogy.medieval, stating that a new solution to Agatha's parentage was forthcoming. Within a few days after the issue containing the article arrived in the mailboxes of subscribers, brief outlines of the main arguments were appearing in the newsgroup, so that a wide circle of genealogists, including many who did not subscribe to the Register, became quickly aware of the new research. Follow-up articles were mentioned on the newsgroup as soon as they appeared, and this no doubt contributed to the momentum. However, one also gets the feeling that the appearance of Jetté's article altered the common impression that the matter had been "settled", and that this has encouraged genealogists to try to find new solutions to the problem.

In 1998, the New England Historical and Genealogical Register published an article by David Faris and Douglas Richardson which severely criticized Jetté's article, and argued that the German Hypothesis was correct. There was no attempt to review the basic evidence, and most of the paper was spent criticizing specific points in Jetté's article. Also in 1998, Norman W. Ingham published two articles strongly supporting the Russian Hypothesis. One, in the New England Historical and Genealogical Register, concentrated mainly on the onomastic evidence relevant to the problem [Ingham (1998a)]. The longer (and more important) article, in Russian History, has given what is to date the most detailed discussion in favor of the Russian Hypothesis and also the most ardent criticism (at times excessive) of the German Hypothesis [Ingham (1998b)]. In 2000, Janko Pavsic supported the Russian Hypothesis and argued against the German Hypothesis, in a paper that concentrated on generally unconvincing onomastic arguments [Pavsic (2000); English translation in Pavsic (2001)]. In 2002, an article by John Carmi Parsons strongly criticised Janko Pavsic's article [Parsons (2002); in a bizarre blunder, the author of the previous paper was misnamed "Pavel Javsic", with the incorrect surname appearing throughout the paper]. The article contains a good discussion of how genealogical arguments involving onomastics are often pressed to far. The author concluded that the case was not proven either way (referring to the German Hypothesis and the Russian Hypothesis), and ended by offering two additional theories "to indicate just how far we are from the last word on this question." [Parsons (2002), 52] One theory suggested that Edward may have married twice. The other suggested that Agatha may have been the daughter of a count Cristinus (the "Cristinus Hypothesis"). Neither of these theories was pressed as being definitive. Also in 2002, Gregory Lauder-Frost published a two-page article pointing out the controversy which had developed regarding Agatha's origins [Lauder-Frost (2002)]. Several of the recently published papers were briefly mentioned, and the author expressed a preference for the Hungarian Hypothesis. In 2003, a long article on Agatha's origins by Ian Mladjov appeared. The article gave a very good outline of the main research on Agatha during the past fifty years, and concluded by offering yet another theory on her origin, here called the "Bulgarian Hypothesis". This hypothesis suggests that Agatha was a daughter of Gavril Radomir (d. 1015), king of Bulgaria, a maternal granddaughter of king Géza I of Hungary (d. 997), and a stepdaughter of Aba Samuel (d. 1044), another king of Hungary. It also makes her a paternal granddaughter of another woman named Agatha [Mladjov (2003)]. This paper is hindered in many places by the appearance of spaces where accented characters should have appeared instead. This does not seem to be the author's fault, but appears to be the result of poor production standards by the publisher in the printing of the article. In 2003, an article by William Humphreys argued in favor of the Russian Hypothesis [Humphreys (2003)]. In a follow-up article in 2004, Humphreys proposed as an alternate hypothesis the possibility that Agatha was a sister of Anastasia, wife of Iaroslav's son Vsevelod (d. 1093), and apparently daughter of the Byzantine emperor Constantine IX "Monomachos" (the "Byzantine Hypothesis") [Humphreys (2004), 280]. However, the author regarded this as less likely than the Russian Hypothesis [ibid., 287]. The main thrust of the article is that Agatha might have had Greek ancestry, for he also suggests that Iaroslav may have been the son of Anna of Byzantium [ibid., 284-5]. In 2009, John P. Ravilious proposed what is here called the "Polish Hypothesis", in which Agatha is conjectured to be a daughter of duke Mieszko II of Poland by his wife Richenza, granddaughter of the emperor Otto II by his Byzantine wife Theophanu [Ravilious (2009)].

The Sources

Here, we give an outline of the main primary sources for the parentage of Agatha, and for the exile of the sons of Eadmund Ironside. There are numerous sources other than the ones which are listed here, but the other sources which are of any value have taken their information from one of the sources given below.

Adam of Bremen

Adam of Bremen, writing about 1070, states that Eadmund (Ironside, here wrongly described as a brother of Ćthelred II) was killed by poison, and that his sons (unnamed) were condemned to exile in Russia ["Frater Adelradi Emund, vir bellicosus, in gratiam victoris veneno sublatus est; filii eius in Ruzziam exilio dampnati." Adam of Bremen, ii, 51, MGH SS 7: 324]. Although he gives no information on Agatha, Adam's account is important in giving early testimony that the exile of Eadmund's sons included time in Russia.

The Anglo-Saxon Chronicle

Accounts in the Anglo-Saxon Chronicle entered under the years 1057 and 1067, but not contemporary with those dates (having been written perhaps soon after 1100, perhaps a bit earlier), are the first to give Agatha's name and information about her origin. The 1057 entry gives Agatha's name and states that she was a relative of the emperor (unnamed) ["Her com Eadward ćţeling to Englalande se wćs Eadwerdes brođor sunu kynges Eadmund 'cing'; Irensid wćs geclypod for his snellscipe. Ţisne ćţeling Cnút hćfde forsend on Ungerland to beswicane. Ac he ţćr geţeh to godan men, swa him God uđe, & him wel gebyrede. swa ţ. he begeat ţćs caseres mága to wife, & bi ţćre fćgerne bearnteam gestrynde, seo wćs Agathes gehaten." ASC(D) s.a. 1057 (Translation: "Here the ćtheling Edward came to England; he was son of King Edward's brother Edmund, [who] was called 'Ironside' for his bravery. King Cnut had sent this ćtheling away into Hungary to betray, but he there grew to be a great man, as God granted him and became him well, so that he won the emperor's relative for wife, and by her bred a fine family; she was called Agatha." ASC(Eng), 187-8)]. The 1067 account states rather vaguely that Margaret's mother's (i.e., Agatha's) family goes back to the emperor Heinrich (which Heinrich is not stated) ["... & hire modor cynn gćđ to Heinrice casere ţe hćfde anwald ofer Rome." ASC(D) s.a. 1067 (p. 202) (Translation: "... and her [i.e., Margaret's] mother's family goes back to the emperor Henry who had dominion over Rome." ASC(Eng), 202)]. Only the Hungarian exile is mentioned.

John ("Florence") of Worcester

According to the chronicle of John of Worcester (1118?, often attributed to "Florence" of Worcester), king Cnut sent the sons of Eadmund, Eadweard and Eadmund, to the king of the Swedes to be killed, but the king of the Swedes instead sent them to Salomon, king of Hungary, who brought them up. Eadmund died, but Eadweard married Agatha, daughter of a "germanus" of the emperor Heinrich ["Dedit etiam consilium Edricus, ut clitunculos, Eadwardum et Eadmundum, regis Eadmundi filios, necaret; sed quia magnum dedecus sibi videbatur ut in Anglia perimerentur, parvo elapso tempore, ad regem Suanorum occidendos misit; qui, licet foedus esset inter eos, precibus illius nullatenus voluit acquiescere, sed illos ad regem Ungariorum, Salomonum nomine, misit nutriendos, vitćque reservandos: quorum unus, scilicet Eadmundus, processu temporis ibidem vitam finivit; Eadwardus vero Agatham, filiam germani imperatoris Heinrici, in matrimonium accepit, ex qua Margaretam Scottorum reginam, et Christinam sanctimonialem virginem, et clitonem Eadgarum suscepit." John Worc. s.a. 1017 (1: 181)]. John's genealogical appendix has the same statement, this time with Heinrich identified as the emperor Heinrich III ["Eadwardus vero Agatham, filiam germani imperatoris Heinrici III., in matrimonium accepit, ex qua Margaretam reginam Scottorum, et Chrisianam virginem, et clitonem Eadgarum, suscepit." John Worc. 1: 275]. It should be noted here that John has misnamed the Hungarian king, as Salomon did not reign until years later. Simeon of Durham has the same statement, obviously copied directly from the work of John ["Eadwardus vero Agatham filiam Germani imperatoris Henrici in matrimonium accepit." Sim. Durh., Hist. Regum, c. 130 (2: 155)]. Another twelfth century source which evidently copied from John is the Chronicle of Melrose [Chron. Melrose, s.a. 1017 (pp. 43-4)].

Note from the capitalization of the entry just quoted that the editor of Simeon of Durham was interpreting the sentence as meaning "Indeed, Eadward took in marriage Agatha, a daughter of the German emperor Heinrich." Such a reading of the word germanus is the origin of the theories that Agatha was a daughter of either Heinrich II or Heinrich III, both known to be false relationships. In fact, the word used here is not "Germanus" the geographical adjective, but "germanus" the relationship term, and the correct translation seems to be "Indeed, Eadward took in marriage Agatha, daughter of a germanus of the emperor Heinrich." The meaning of "germanus" is discussed in Appendix 3.

William of Malmesbury

William of Malmesbury, in his Gesta Regum Anglorum, written in 1125, states that Eadmund Ironside's sons Eadwig [Edwius, a mistake for Eadmund] and Eadweard were sent to the king of Sweden to be killed, but that being spared by his mercy, they went to the king of Hungary, where the elder died and the younger brother (presumably Eadweard) married Agatha, sister of the queen ["Filii ejus Edwius et Edwardus, missi ad regem Swevorum ut perimerentur, sed miseratione ejus conservati, Hunorum regem petierunt; ubi, dum benigne aliquo tempore habiti essent, major diem obiit, minor Agatham reginć sororem in matrimonium accepit." Wm. Malmes., Gesta Regum, c. 180 (vol. 1, p. 218)]. Here, Eadweard's brother is incorrectly called Eadwig.

Orderic Vitalis

Orderic Vitalis, writing between 1124 and 1142, states that Cnut sent the boys Eadmund and Eadweard to Denmark, and ordered his brother Svend [sic] king of Denmark (presumably a mistake for Harald, king of Denmark) to kill them, but that he sent the boys to the king of the "Huns" (i.e., Hungarians), where Eadmund died an untimely death. He then states that Eadweard married a daughter of the king of the Hungarians, and that Eadweard himself reigned over the Hungarians ["Eduardum vero et Edmundum filios Edmundi, elegantes albeolos, in Daciam relegavit, et Sueno regi Danorum fratri suo, ut eos interficeret, mandavit. At ille generosos et innocentes pueros nequiter necare contempsit, sed orta occasione regi Hunorum illos quasi nepotes suos obsides dedit. Ibi Edmundus clito immatura morte obiit. Eduardus vero Dei nutu filiam regis in matrimonium accepit, et super Hunos regnavit. Edgarum vero Adelinum, et Margaritam reginam Scotorum, et Christianam sanctimonialem genuit; ..." OV, i (vol. 1, p. 178)]. On another occasion Orderic names Eadweard's father-in-law as king Salomon, a chronological impossibility ["Haec [Margarita] nimirum filia fuit Eduardi, regis Hunorum, qui fuit filius Edmundi cognomento Irnesidć, fratris Eduardi regis Anglorum, et exul conjugem accepit cum regno filiam Salomonis regis Hunorum." ibid., viii, 22 (vol. 3, p. 398)].

Laws of Edward the Confessor

The so-called Laws of Edward the Confessor, actually dating from Norman times, 1115×50 (probably 1130×5), state that Eadweard took refuge in Russia, that he was decently retained there by the Russian king Malesclodus, and that he was married there to a wife of noble descent ["Iste supradictus Eadmundus habuit filium quendam, qui uocatus est Eadwardus. Qui, mortuo patre, timore regis Cnuti aufugit de ista terra usque ad terram Rugorum, quam nos uocamus Russeiam. Quem rex ipsius terre, Malesclodus nomine, ut audiuit et intellexit, quis et unde esset, honeste retinuit eum. Et ipse Ćdwardus accepit ibi uxorem ex nobili genere, de qua ortus est ei Eadgarus ađeling et Margareta regina Scotie et Cristina soror eius." Laws Edw. Conf., c. 35-35.1 (p. 664)]. This passage was also contained in the chronicle of Roger de Hoveden, where some manuscripts read "... ad regnum Dogorum, quod nos melius vocamus Russiam. Quem rex terrć Malescoldus nomine, ..." [Rog. Hoveden, Legal Appendix to Chronica, 2: 236]. The identity of Malesclodus is discussed in Appendix 6. An interpolation following "Cristina soror eius" in the above passage, dated about 1210 [Liebermann (1903), xxxiv], states that Margaret's mother was in origin and blood of the kings of the Russians ["Fuitque predicta Margareta generosa ualde et optima, scilicet ex parte patris ex nobili genere et sanguine regum Anglorum-Britonum, ex parte uero matris ex genere et sanguine regum Rugorum; ..." Laws Edw. Conf., c. 35.1*** (p. 664)]. This is apparently intended to imply that Margaret's mother was a daughter of Malesclodus, although that is not explicitly stated. This interpolation does not occur in Roger de Hoveden's work.

Geoffrey Gaimar

Geoffrey Gaimar wrote his Lestorie des Engles probably after 1135 and certainly before 1147. Geoffrey has a long legendary account of the sons of Eadmund Ironside, whom he incorrectly calls Eadgar and Ćthelred ["Li vns ert Edgar apelez, / Li altres out nun Edelret:" ("One was called Eadgar, / The other's name was Ćthelred") Gaimar 4516-7]. Of these, it is Eadgar who is later stated to be the father of Margaret and Eadgar the Ćtheling [Gaimar 4647-4652], so it is evidently he who is intended to be identified with Eadweard the Exile. According to Geoffrey's story, the two boys were entrusted to a Dane named Walgar, and sent to Denmark, where they remained for twelve years [Gaimar 4503-4522]. After this time, Cnut heard from his wife Emma that the English wanted to make the boys king, so Cnut ordered that they should be maimed [Gaimar 4567-70]. Walgar, warned of the plot, fled with the boys to Hungary, passing through Russia on the way ["Si espleita son errer. / Ken sul cinc iurs passat Susie, / E vint en terre de Hungrie. / Le siste iur est ariuez / De suz Gardimbre, la citez: / Li reis i ert e la raine, / A ki Hungrie estait acline." ("He [Walgar] so well accomplished his journey / That in only five days he passed Russia, / And came to the land of Hungary. / The sixth day he arrived / Beneath the city of Gardimbre. / The king was there and the queen, / To whom Hungary was subject.") Gaimar 4582-8]. Walgar then entrusted the boys to the king of Hungary, telling him that they were the rightful heirs of England [Gaimar 4592-4618]. After three years "Eadgar" became a lover of the daughter of the king (presumably of Hungary) and the lady became pregnant ["Edgar out nun, mult fu senez. / La fille al rei en fist son dru; / E cil lamat, co fu seu: / Ainz ke passast tut lan enter, / Avint la dame a enceinter." ("Eadgar was his name. He was well favoured. / The king's daughter took him for her lover. / And he loved her; this was known; / Before a whole year had passed, / The lady became pregnant.") Gaimar 4624-8]. Then the king gave his daughter to "Eadgar" and made him his heir ["Li reis sa fille a Edgar donat: / Veanz sa gent cil lespusat; / E li reis fist a tuz sauer, / Apres son iur sait Edgar heir:" ("The king gave his daughter to Eadgar. / Before his people, he married her, / And the king gave all to know / That Eadgar should be his heir after his days.") Gaimar 4639-42]. "Eadgar" and his wife become the parents of Margaret and Eadgar the Ćtheling ["De cest Edgar e de sa femme, / Eissit la preciose gemme, / Margarete lapelat lom, / Raine en fist rei Malcolom. / Ele aueit vn son frere ainnez, / Edgar lAdeling estait nomez." ("From this Eadgar and his wife / Issued the precious gem, / Margaret they called her. / King Malcolm made her his queen. / She had an elder brother, / Eadgar the Ćtheling was he named.") Gaimar 4646-4652]. Then, after the death of their father, Margaret and Eadgar the Ćtheling are sent for by the English, but on the way there a storm drives them to Scotland, where Malcolm seizes them and marries Margaret [Gaimar 4657-4662]. Although Geoffrey clearly used sources which are now lost, his work has such large doses of legend and romance that he is not a trustworthy source. His account of the exile of Eadmund Ironside's sons clearly has a large dose of fiction, and any attempt to disentangle truth from Geoffrey's fictionalized account in order to use them as "evidence" is fraught with difficulties.

Ailred of Rievaulx

The Genealogia Regum Anglorum was written in 1153 or 1154 by Ailred of Rievaulx, who was brought up in the Scottish royal household and got some of his information directly from king David, grandson of Agatha [Ingham (1998), 249]. Unfortunately, the information he gives is contradictory. On one occasion he states that Agatha's daughter Margaret was descended from the royal seed of the English and the Hungarians ["Hanc religiosa regina Margareta, hujus regis mater, quć de semine regio Anglorum et Hungariorum exstitit oriunda, allatam in Scotia quasi munus hćreditarium transmisit ad filios." Ailred of Rievaulx, Genealogia Regum Anglorum, PL 195: 715]. Thus, in this statement, he appears to be implying that Agatha was of Hungarian royal descent.

Ailred's next statement not only contradicts the previous statement, but gives a marriage to Eadweard's brother Eadmund, stating that the Hungarian king gave his daughter to Eadmund as a wife ["At puerulos filios Edmundi ferire metuens prć pudore, ad regem Suavorum eos interficiendos transmisit. Rex vero Suavorum nobilium puerorum miseratus ćrumnam, ad Hungariorum regem eos destinat nutriendos. Quos ipse benigne accepit, benignius fovit, benignissime sibi in filios adoptavit. Porro Edmundo filiam suam dedit uxorem; Edwardo filiam germani sui Henrici imperatoris in matrimonium junxit. Sed paulo post Edmundus de temporalibus ad ćterna transfertur: Edwardus sospitate et prosperitate fruitur." Ailred of Rievaulx, Genealogia Regum Anglorum, PL 195: 733] The awkward sentence giving the marriage of Eadweard ("Edwardo filiam germani sui Henrici imperatoris in matrimonium junxit.") would appear to translate as "He [i.e., the Hungarian king] joined the daughter of his 'germanus' the emperor Heinrich in marriage to Eadweard." A way out of this awkwardness is provided by some marginal notes which appear in the Hengwrt MS. version of Henry of Huntingdon's Historić Anglorum, in a hand of ca. 1200, which were published by Arnold in his edition of Henry's work. These passages are obviously taken from a version of Ailred's work, and the one relevant here reads as follows: "At puerulos, scilicet filios Edmundi, ferire metuens pro pudore, ad regem Swanorum eos interficiendos transmisit. Rex vero Swanorum nobilium puerorum miseratus ćrumnam ad Hungariorum eos regem destinavit nutriendos. Quos ipse benigne susceptos benignius fovit, benignissime sibi in filios adoptavit. Porro Edmundo filiam suam dedit uxorem, Edwardo filiam germani Henrici imperatoris in matrimonium junxit. Sed paulo post Edmundus de temporalibus ad ćterna transfertur; Edwardus sospitate et prosperitate perfruitur." [H. Hunt., p. 296 (the first appearance of "ad" appears in the printed edition as "sd", apparently a printing error)] The absence of the word "sui" then leads to the more natural translation "He joined the daughter of a 'germanus' of the emperor Heinrich in marriage to Eadweard." This reading is confirmed by another statement of Ailred in the Genealogia ["... Edwardum cum uxore sua Agatha germani sui filia liberisque ejus, Edgaro Edeling, Margareta atque Christina, ..." ("sui" refers to the emperor) Ailred of Rievaulx, Genealogia Regum Anglorum, PL 195: 734], as well as a vaguer statement by Ailred in his Life of Edward the Confessor, which states that the Roman emperor gave his relative ("cognata") in marriage to the son of Eadmund Ironside ["Imperator Romanus cujus cognatam regis nepos filius Eadmundi ferrei lateris, unus e duobus quos exsilio Cnute damnaverat, uxorem duxit, ..." Ailred, Vita S. Edwardi Regis, PL 195: 745].

The Theories

The main hypotheses are listed here, along with the labels that they have been assigned for purposes of the discussion below. A few impossible theories which can be easily dismissed are not given labels.

The German Hypothesis (main version):
Conjectured father (possible): Liudolf, d. 15 or 23 April 1038, count (Braunschweig).
Conjectured mother (possible): Gertrude.

The Russian Hypothesis:
Conjectured father (possible): Iaroslav I, d. 1054, grand prince of Kiev.
Conjectured mother (possible): Ingegerd, daughter of Olaf, king of Sweden.

The Polish Hypothesis:
Conjectured father (improbable): Mieszko II Lambert, d. 10 May 1034, king of Poland.
Conjectured mother (improbable): Richenza, daughter of Ezzo, count palatine of Lorraine.

The Bulgarian Hypothesis:
Conjectured father (improbable): Gavril Radomir, d. 1015, emperor of Bulgaria.
Conjectured mother (improbable): NN, sister of István (Stephen) I, king of Hungary.

The Hungarian Hypothesis:
Conjectured father (very improbable): István (Stephen) I, d. 1038, king of Hungary.
Conjectured mother (improbable): Gisela, sister of Heinrich II, emperor.

The Cristinus Hypothesis:
Conjectured father (very improbable): Christinus, count.
Conjectured mother (highly improbable): Oda, daughter of Bernhard, count of Haldensleben.

The German Hypothesis (alternate version):
Conjectured father (very improbable): Ernst II, d. 17 August 1030, duke of Swabia.

The Bruno Hypothesis:
Conjectured father (very improbable): Bruno, d. 24 April 1029, bishop of Augsburg, 1007-1029, brother of emperor Heinrich II.

The Byzantine Hypothesis:
Conjectured father (no reasonable basis): Constantine IX "Monomachos", d. 1055, Byzantine emperor.

Falsely attributed father: Salomon, d. 1087, king of Hungary, 1063-74.
[OV; see above for details] Orderic Vitalis is the only early medieval source to name the alleged father of Eadweard's wife. However, the claim is chronologically impossible.

Falsely attributed father: Heinrich II, d. 13 July 1024, emperor.
[e.g., Burke (1848-51), 1: ped. cxix; 2: ped. xxxviii] Although the secondary sources giving this relationship of which I am aware do not state sources, it is clear that this theory came about because "filia germani imperatoris Heinrici" was misinterpreted as "daughter of the German emperor Heinrich" [see Appendix 3 for the meaning of germanus]. However, it is virtually impossible that a child of Heinrich II, if one had existed, would have gone unmentioned by continental sources.

Falsely attributed father: Heinrich III, d. 5 October 1056, emperor.
[e.g., Baverstock (1832), 20] This comes about by the same misunderstanding as the previous theory. It is chronologically impossible.

Falsely attributed father: Hardicanute, d. 8 June 1042, king of Denmark and England.
[Felch (1894), 2, mentions this claim, the ultimate source of which was apparently royal pedigrees published by Reusner in 1592; this reference was pointed out by Todd Farmerie on soc.genealogy.medieval] The supposed logic behind this chronologically impossible theory is unknown.

Background: the exile of the sons of Eadmund Ironside and the return of Eadweard the Exile

It is clear that Eadmund Ironside's infant sons, Eadmund and Eadweard, went into exile soon after his death in 1016. It is also clear that the surviving son, Eadweard, was living in Hungary when his uncle king Eadweard the Confessor sent for him in the 1050's. Eadweard's activities between 1016 and 1057 are poorly documented, and depend almost entirely on sources which are not contemporary. While it is likely that our sources from the first half of the twelfth century preserve some reliable traditions, they are, as we can see, not entirely consistent, and in some cases verifiably false.

Adam of Bremen is the only source to mention the princes which is close to being contemporary. He states that they were condemned to exile in Russia. Thus, there is a strong probability that at least part of their time in exile was spent in Russia. The Anglo-Saxon Chronicle knows only about the time in Hungary. John of Worcester, William of Malmesbury, and Ailred of Rievaulx all state that the exile started off in Sweden and that they then went to Hungary. Orderic Vitalis and Geoffrey Gaimar have them going first to Denmark and then to Hungary, with Gaimar mentioning that they passed Russia on the way to Hungary. The Laws of Edward of the Confessor mention only Eadweard, and states that he went to Russia, where he was married. The modern consensus seems to be an itinerary which includes Sweden, Russia, and Hungary, in that order.

It has frequently been stated that Eadmund and Eadweard went from Sweden to Russia in 1028 and from Russia to Hungary in 1046, but it is very misleading to suggest that such chronological details can be deduced from the slim evidence at hand, even if the itinerary of Sweden to Russia to Hungary is tentatively accepted. Both dates come primarily from pieces of evidence that may not have anything to do with the movements of the exiled family. The 1028 date is primarily based on the fact that king Cnut conquered Norway in that year [ASC(E) s.a. 1028]. Olaf of Norway and his son were forced to flee Norway, and they went to Sweden and then to Russia [Óláfs saga Helga, c. 181, Heimskringla, 474]. Jetté has Cnut defeating the Swedes (is this an error for the Norwegians?) in 1028 [Jetté (1996), 418; he is followed by Ingham (1998b), 234]. It has been conjectured that it was at this time that the English princes went from Sweden to Russia [Vajay (1962), 72; Wunder (1975), 82; Jetté (1996), 418; Ingham (1998b), 234]. Ronay accepts the accounts of Geoffrey Gaimar and Orderic Vitalis that the princes were first sent to Denmark [Ronay (1989), 28-40 passim], and he then has the English princes move from Denmark to Sweden and then from there to Russia in 1028×9, following the path of Olaf [ibid., 40-1, 52-3]. After spending many years in Russia, Eadweard is then supposed to have moved to Hungary in 1046 as a part of the army that helped Andrew gain the Hungarian throne in that year [Vajay (1962), 72-3; Wunder (1975), 82; Ronay (1984), 47; Jetté (1996), 419-20; Ingham (1998b), 235].

The problem with these scenarios comes from the apparent underlying assumption that the movements of the exiled family must necessarily be the direct result of political events which appear in the surviving sources. It is misleading to take such attempts at "reading between the lines" and interpret them as verified history. In fact, our information on the exile is very fragmentary and comes almost entirely from sources of a century later. It can be regarded as reasonably certain that the exiled family was in Hungary at the time that king Eadweard the Confessor sent for them. It is also probable that their exile included time in Sweden and Russia (but see the Polish Hypothesis below for a different opinion on this). However, attempts to narrow down the time of movement from one region to another are only conjectures.

Comparing the sources

Clearly, the different sources say different things about the origin of Agatha. The early accounts we have on Agatha's parentage can be placed in four main categories.

The natural first attempt would be to look for an individual who simultaneously fits into all of the above categories. However, it has not proved possible to find a parentage for Agatha which agrees with all of these sources. Thus, there seems to be the inescapable conclusion that some of these sources are unreliable. The following table indicates how well the various pieces of basic information match with the various hypotheses which have been proposed, ranging from an excellent fit with the statement of the evidence to being inconsistent with the evidence. Some cases which would involve significantly stretching the definition of a word have been rated poor or very poor. Of course, the first two columns are related, but one is more specific than the other. The first two columns assume that any emperor Heinrich is allowed.

  Relative
of emperor
Heinrich
Daughter
of
germanus
of emperor
Heinrich
Sister of
Hungarian
queen
Daughter of
Hungarian
king
Descendant
of Russian
Royaly
Bruno Hypothesis excellent excellent poor inconsistent inconsistent
Bulgarian Hypothesis very poor inconsistent inconsistent very good inconsistent
Byzantine Hypothesis inconsistent inconsistent inconsistent inconsistent inconsistent
Cristinus Hypothesis poor inconsistent inconsistent inconsistent good
German Hypothesis excellent very good poor inconsistent inconsistent
Hungarian Hypothesis excellent poor inconsistent excellent inconsistent
Polish Hypothesis good inconsistent excellent inconsistent poor
Russian Hypothesis inconsistent inconsistent excellent inconsistent excellent

As can be seen from the table, every one of the hypotheses rates as inconsistent in at least two columns, and each theory rates as either inconsistent or poor in at least three columns. Clearly, it has not been possible to find a theory which fits well with all of the basic primary evidence. It would be unwise to try to assign a "score" to each of the theories by somehow tabulating the results from this table, which is only a rough guide (and contains some entries which are judgement calls). The evidence for Agatha's parentage is complicated, and depends on many additional factors which could not be easily enumerated on such tables, and which may be weighed differently by different researchers. Also, the information in the table could be potentially misleading. Opponents of the German Hypothesis could complain that the information from the first two cloumns might not be independent, and that including both gives the German Hypothesis an extra "vote". Opponents of the Russian Hypothesis could complain that the last column is based on a late interpolation. Thus, it is important to consider the comparative reliability of each of the sources.

Let us first consider the accounts in which Agatha is a member of the Hungarian royal family. As already noted, there are three basic twelfth century sources for this. Orderic Vitalis states that Eadweard married a daughter of king Salomon of Hungary and then became king of Hungary. (It would appear that several researchers on Agatha have noticed only the statement in which Orderic has Eadweard marrying a daughter of an unidentified king of Hungary, and have overlooked another passage in Orderic's work where the father-in-law is identified as Salomon.) However, it is clearly chronologically impossible for Salomon to be the father of Agatha (see Appendix 5), and the statement that Eadweard was king of Hungary is also false. Thus, Orderic is clearly not a reliable guide to the identity of Agatha's father. The second source making Agatha a Hungarian princess is Geoffrey Gaimar. As already noted, he is not a reliable source, and there is no reason to trust his testimony unless it is confirmed elsewhere. This leaves the account of Ailred of Rievaulx, who states that St. Margaret was of royal English and Hungarian descent ["Hanc religiosa regina Margareta, hujus regis mater, quć de semine regio Anglorum et Hungariorum exstitit oriunda, ..." Ailred of Rievaulx, Genealogia Regum Anglorum, PL 195: 715]. Now, Ailred personally knew king David I, Marageret's youngest son, and his testimony would therefore ordinarily rank highly, were it not for the fact that he later contradicts himself by distinguishing Eadweard's wife from the Hungarian king's daughter, whom he marries to Eadweard's brother Eadmund (see Appendix 2). Such a marriage of Eadmund, if true, would certainly provide a convenient explanation for the contradictory attribution of Agatha as a Hungarian king's daughter, which would then be explained as an error. At the very least, it shows that the evidence for making Agatha a daughter of the Hungarian king is weak.

The statement that St. Margaret was "ex parte uero matris ex genere et sanguine regum Rugorum" comes from an interpolation made about 1200 to the Laws of Edward the Confessor [" Laws Edw. Conf., c. 35.1*** (p. 664)]. What we would like to know is whether the interpolator was taking this information from some other source to which he had access, or whether he was simply expanding on the earlier statement that Eadweard was married in Russia to a woman of noble descent. The latter is much more likely, as Norman Ingham seems to concede, while still arguing that the interpolation be accepted into evidence [Ingham (1998b), 256]. However, the interpolator's statement would carry significant weight only if he were working from some other source, which is much less likely. Thus, the main value in the Laws of Edward the Confessor lies in the uninterpolated part, which makes only the vague statement that Agatha was of noble descent.

This leaves the statement of John of Worcester and others that Agatha was a daughter of a germanus of emperor Heinrich (along with the Anglo-Saxon Chronicle's vaguer statement along the same lines) and the statement of William of Malmesbury that Agatha was a sister of the Hungarian queen. Unless there are emperor's brothers or Hungarian queens out there who have remained undocumented, trying to get a scenario which satisfies both William of Malmesbury and John of Worcester does not look promising. Among the current proposals, the closest would seem to be the German Hypothesis, if William of Malmesbury's soror is interpreted as meaning "first cousin". However, the use of soror in this way is very rare, and it is much more likely that William of Malmesbury meant soror in the usual sense of sister. Thus, we are left with the alternative that one of the two sources is mistaken with regard to Agatha's origins.

So, how reliable are John of Worcester and William of Malmesbury? John and William had very different styles. William was a historian who interpreted his sources and rewrote the account in his own words. John was a chronicler who was more likely to pass information along in the same form in which he found it in his sources. Both are highly regarded, but neither would meet modern standards, and both made errors. In fact, they each made a similar error in their accounts of the exile of the princes, in the passages leading up to their statements of Agatha's origin. John of Worcester gives the wrong name to the king of Hungary who accepted the princes in exile, calling him Salomon, a king who did not reign until later. William of Malmesbury gives the wrong name to one of the exile brothers, calling him Eadwig (the name of an uncle) instead of Eadmund. Neither of these errors seems sufficient to reject the statements about Agatha without further evidence. After praising William of Malmesbury as a historian, Norman Ingham says of John of Worcester: "The form of the passage in John reveals his cut-and-paste method of composition; he has patched several pieces of 'information' together without proper transitions - Sweden, King Salomon of Hungary, the death of Edmund, the marriage of Edward - thereby making jumps in time and logical coherence." [Ingham (1998b), 248] However, there seems to be a double standard here. After all, with less transition than John of Worcester, William of Malmesbury jumps from Sweden to the (unnamed) king of Hungary to the death Eadmund to the marriage of Eadweard. Indeed, if it were not for the different information on Agatha's parentage, there would be reason to suspect that John and William were using the same source here.

William of Malmesbury's statement that Agatha was a sister of the queen of Hungary is not confirmed by any independent source. Besides sources which are clearly dependent of John of Worcester, statements that agree with John's account appear in both the Anglo-Saxon Chronicle and Ailred of Rievaulx. Are these statements independent, or do they go back to a common source? If so, can anything be determined about this source? The fact that John of Worcester used a source very similar to the D (Worcester) manuscript of ASC is well known [See the remarks by Plummer in ASC 2: lxxxiii-lxxxiv]. However, the problem with assuming that John used D or a source similar to D for his statement is that John gives more information, stating that Agatha was a daughter of a germanus of emperor Heinrich, while D only states that she was a relative. Thus, since D (being the earlier source) did not copy from John, and it is improbable that John invented the more precise additional information, the connection of both John and the D manuscript to Worcester would indicate that they both got their statements from a common source located there, as suggested by Moriarty [Moriarty (1952), 55]. Ealdred, later archbishop of York, who was sent on the mission to bring Eadweard the Exile back to England, had also been bishop of Worcester, so Worcester would be a natural place to find information about Eadweard's family. With Ailred the situation is not so clear. He certainly had at least one independent source which led him to his statements that Margaret's mother had royal Hungarian blood and that Eadmund married a daughter of the Hungarian king. Ailred must have had another source which led him to his statements in three separate places that Agatha was a daughter of a germanus of emperor Heinrich or a cognata of the emperor. This could very well have been the same Worcester source used by John of Worcester and the D manuscript of the Chronicle. Indeed, Ailred's use of the word germanus here makes this more likely than not. Thus, for convenience, let us use the term "Worcester Source" to denote this hypothetical common source of the statement that Agatha was a daughter of a germanus of emperor Heinrich.

If we assume that there was a common source behind all of these statements that Agatha was a relative of (or a daughter of a germanus of) emperor Heinrich, then what can we say about the date of such a source? It would certainly have to predate the writing of the 1057 and 1067 entries into the D manuscript of the Chronicle. While it is clear from the wording itself that the entries are not strictly contemporary, there has been some disagreement about how much later they were written. Because Margaret's ancestry from the house of Wessex is given, Plummer thought that the entry was not written until after the marriage of Margaret's daughter to Henry I in 1100 [ASC 2: lxxviii]. On the other hand, N. R. Ker dated the handwriting of the 1071-9 entries (written after the 1057 and 1067 entries) to the 1070's or 1080's [Ker (1957), 254]. Dorothy Whitelock expressed an opinion somewhere in between. She stated that Plummer's dating to after 1100 was not certain, but she pointed out that no life of Margaret would have been written before 1093, and suggested that Ker's dates seemed too early [Whitelock (1979), 1: 115]. Here, Whitelock was assuming that the information about Margaret came from a written source (presumably a life written after her death), but G. P. Cubbin, editor of the most recent edition of the Worcester (D) manuscript of the Anglo-Saxon Chronicle, stated that a formal source would not be needed, and that Ker's dating "need not be doubted on these grounds" [ASC-D (Cubbin), lxxiv]. Thus, since the dating of the handwriting makes a date significantly later than 1100 doubtful, and since Plummer's reasons are far from conclusive anyway, our hypothetical Worcester Source can be dated to ca. 1100 or earlier.

One common way of resolving a discrepancy between two sources is to explain how one of them arose as a natural error. One such explanation, first given by Jetté and expanded by Ingham, claims that the Worcester Source was an error resulting from a misunderstanding of the statement that Agatha was a sister of the Hungarian queen [Jetté (1996), 423; Ingham (1998b), 244, 248]. This needs to be examined in detail. The argument assumes that a statement that Agatha was a sister of the queen of Hungary was the main surviving piece of information available to the chroniclers. Two Hungarian queens during the period were Gisela, wife of Stephen I and sister of emperor Heinrich II, and Judith, wife of Salomon and daughter of emperor Heinrich III. While it is chronologically impossible for either of these queens to have been a sister of Agatha, the theory goes that one or the other of them was mistaken for the sister of Agatha (starting with an earlier source similar to William of Malmesbury which stated that Agatha was a sister of the Hungarian queen), and as a result it was mistakenly assumed that Agatha was related to an emperor Heinrich, resulting in the information given in the Worcester Source. With the sources depending on the Worcester Source explained away in this manner, William of Malmesbury is portrayed as the only early reliable source. However, there is a serious problem with this scenario. If the author of the Worcester Source mistakenly assumed that Agatha was a sister of either Gisela or Judith, then he would have deduced that Agatha was a daughter or a sister of an emperor Heinrich. Why then, would the Worcester author make Agatha the daughter of a germanus of Heinrich, contrary to this deduction? Having made a mistake identifying the queen, he would then have had to make another mistake regarding the relationship of that queen to the emperor in order to get the account that was passed on. Thus, trying to explain away the information of the Worcester Source as a simple error from an account similar to William of Malmesbury simply will not work. Looking at the other direction, if Agatha were a daughter of a germanus of Heinrich III, as the genealogical appendix of John of Worcester states, then Agatha would be a first cousin of Salomon's wife Judith. Thus, the alternate argument would be that the Worcester Source was right, and that William of Malmesbury, or his source, erred by changing a cousin into a sister. This may not be right, but it would be more likely than the more complicated series of two errors needed to make Jetté's theory true.

Thus, of the sources giving information about the ancestry of Agatha, the accounts giving her Hungarian royal blood seem very doubtful, and the one giving her royal Russian blood is a late interpolation. If the accounts making Agatha a relative of the emperor were independent, their testimony would be fomidable, but it is much more likely that they are not independent, and go back to a hypothetical common source which we have named the "Worcester Source". This Worcester Source would be earlier than the work of William of Malmesbury, and would have to be given a slight edge for that reason. However, no firm conclusion is possible, and either one of the Worcester Source or William of Malmesbury might be correct.

Different theories on the origin of Agatha

As already noted, there is a plethora of hypotheses regarding the origin of Agatha, some of which have first appeared relatively recently. Except for a handful of old theories which can easily be eliminated as impossible, these hypotheses have each been supplied above with a label for ease of reference. The two hypotheses which seem to have the best chance of being true, the German Hypothesis and the Russian Hypothesis, will first be discussed in detail (with some discussions deferred to the Appendices), and then the other theories will be more briefly discussed in the approximate order in which they were proposed.

The German Hypothesis

The German Hypothesis argues that Agatha was a daughter of one of the half-brothers of the emperor Heinrich III, and was apparently first proposed in 1939 by József Herzog [Herzog (1939), in Hungarian, which I cannot read]. The next year, Marcellus von Redlich wrote a short paper accepting Herzog's arguments, and favoring Ernst II of Swabia as the father (the "alternate" version of the German Hypothesis). In 1962, Szabolcs de Vajay published the "main" version of the German Hypothesis, proposing that Liudolf of Braunschweig (Brunswick) was Agatha's father. Since that time, the German Hypothesis has been accepted as proven by many sources, most notably Ronay in his 1984 paper and 1989 book [Ronay (1984, 1989)]. Since the appearance of Jetté's article in 1996 supporting the Russian Hypothesis, the main paper arguing in favor of the German Hypothesis has been the article of Faris and Richardson [Faris-Richardson (1998)].

The logic behind the German Hypothesis is pretty straightforward. In what is probably the earliest source to mention the origin of Agatha, the Anglo-Saxon Chronicle states that she was a relative of the emperor Heinrich. Later, John of Worcester and Ailred of Rievaulx (in one of his contradictory statements) add more detail to this by stating that Agatha was the daughter of a germanus of the emperor Heinrich, and John's genealogical appendix adds even more detail by specifying that the Heinrich in question was Heinrich III. Since Heinrch III had no full brothers, it is then argued by process of elimination that Agatha must have been a daughter of one of the older half-brothers of Heinrich. Consideration of these leads the supporters of main version of the German Hypothesis to conclude that Agatha's father was Liudolf von Braunschweig (d. 1038). No claim has been made that there is any direct evidence making Agatha a daughter of Liudolf.

The discussion of the family of Heinrich III is inadequate in those papers which espouse the German Hypothesis [Herzog (1939); Vajay (1962); Ronay (1984); Ronay (1985); Faris & Richardson (1998)]. However, a detailed discussion of the maternal half-brothers of Heinrich can be found in Appendix 4, based mainly on publications which do not mention Agatha. Evidence is given there that of the three half-brothers of Heinrich, only Liudolf, who is known to have had sons, makes a reasonable candidate for the father of Agatha (the "main" version of the German Hypothesis). The "alternate" version of the German Hypothesis is that another of the half-brothers, Ernst of Swabia, was the father, but, as shown in Appendix 4, he appears to have died without issue. The third half-brother, Hermann of Swabia, being underage in 1030, makes an extremely improbable father for Agatha. Although it is impossible to be sure in such a thinly documented age, it is rather unlikely that any additional half-brothers of Heinrich III have remained unidentified (or at least none who survived long enough to be the father of Agatha).

It is certainly the case that the German Hypothesis accepts some of the primary accounts at the expense of the others, a drawback which it shares with all of its competitors. However, if Eadweard's brother Eadmund were married to a Hungarian princess, as a statement of Ailred suggests (see Appendix 2), then the accounts in which Agatha is Hungarian could be explained as an error due to confusion. The German Hypothesis does not make Agatha a sister of the queen of Hungary, as William of Malmesbury states, but as Vajay pointed out, it does make her a first cousin of Judith/Sophia, wife of king Salomon of Hungary and daughter of Heinrich III [Vajay (1962), 74].

The principle objections, to some extent overlapping, which have been made to the German Hypothesis by its critics are as follows:

Some of these criticisms are valid, but others are overstated or misleading. These objections will be discussed one-by-one.

The German Hypothesis has been criticized because it uses the process of elimination to arrive at a conclusion [Ingham (1998b), 257ff.]. It is certainly true in general that an argument by process of elimination is less desirable than an argument using direct evidence. Indeed, the lack of direct evidence that Agatha was the daughter of Liudolf is one of the weaknesses of the German Hypothesis. However, when the basic information is that Agatha was a daughter of a germanus of Heinrich III, it is likely that our search for a solution is going to involve a search using the process of elimination. As pointed out by Ingham, "(s)uccessful process of elimination requires that we have correctly defined the object of our search, that our sources are exhaustive and we have cast our net widely enough, and that we have good and sufficient criteria for eliminating possibilities." [Ingham (1998b), 257] Thus, we should find all of the germani of Heinrich III an examine them for the likelihood that they could be Agatha's father. If "brother" is the appropriate definition of germanus in this case, as it seems to be, then the number of candidates is small (see Appendix 4). The possibility that there was another sibling of Heinrich who has escaped all of the records is remote. The reasons for excluding Ernst and Hermann as possible fathers for Agatha are given in Appendix 4, and these reasons are strong, if not airtight. I have here assumed that Heinrich III was the emperor intended [as is explicitly stated in John of Worcester's genealogical appendix, John Worc. 1: 275]. The possibility of Heinrich II will be ruled out below in the discussion of the Bruno Hypothesis.

The German Hypothesis has been criticized for the way in which the word germanus is used [Ingham (1998b), 258-60]. Clearly, the meaning of the word germanus is very important to this argument. The strict Latin definition is "full brother", i.e., a brother with both parents in common (as opposed to "half-brother": only one parent in common). An underlying assumption of the German Hypothesis (not clearly stated by its proponents) is that the term germanus allows the definition of half-brother, but also that it is no so loose as to just mean "relative". The usage of John of Worcester with regard to the word germanus (and its corresponding feminine form germana) is examined in detail in Appendix 3. As used by John, the word almost always denoted a sibling (one doubtful exception in more than 50 examples), usually a full sibling in those cases when the information about both parents is known, but sometimes only a half-sibling (two verified cases out of more than 50). Since Heinrich III appears to have had no full brothers (and since any such otherwise unknown full brother would be too young to be the father of Agatha), it seems highly probable that in this case a looser form of germanus was intended. But how loose? If germanus just meant "relative" in this case, why would Agatha be called a "daughter of a relative of Heinrich" instead of just being called a "relative of Heinrich"? The fact that she was called daughter of a germanus clearly indicates that the word germanus was intended in this case to be specific rather than general. Thus, if the information that Agatha was "filia germani imperatoris Heinrici" is true at all (and there is room for argument on that point), then germanus was probably intended to mean "male with at least one parent in common". For further discussion, see Appendix 3.

Of course, as the critics of the German Hypothesis would argue, the claim that Agatha was a daughter of a germanus of Heinrich III is only one of several possibilities given by the sources, so there is no guarantee that this underlying assumption of the German Hypothesis is correct. This disagreement of sources is a definite weakness of the German Hypothesis, a weakness shared with all of the other hypotheses. However, as noted above, critics of the hypothesis have gone further and suggested that the accounts of the Anglo-Saxon Chronicle and John of Worcester have a simple explanation as a blunder. However, it has been shown above that this explanation is not so simple. In particular, why would John of Worcester (or his hypothetical Worcester Source), having supposedly deduced (from a statement similar to that of William of Malmesbury) that Agatha was a sister or daughter of an emperor Heinrich, give the specific information that Agatha was a daughter of a germanus of Heinrich, in direct contradiction to his alleged deduction? The evidence of the Anglo-Saxon Chronicle and John of Worcester (and one of Ailred's accounts) is open to the legitimate objection that contrary testimony exists, but the evididence cannot be explained away in the way that the supporters of the Russian Hypothesis would like.

Jetté made chronological objections to the suggestion that Liudolf was the father of Agatha: "However, the solution would require an exceptionally tight chronology, if Gisela were born in 993 as some have concluded and Agatha, proposed as Gisela's granddaughter, in about 1025." [Jetté (1996), 422, apparently using the date of 1025 given by Vajay (1962), 73, but note that Jetté himself had placed Agatha's birth as late as 1030 on p. 420]. However, as noted above, Agatha could have been born as much as a decade later (although 1030 would be more comfortable), and, as discussed in detail in Appendix 4, Liudolf was probably born before 1010.

Opponents of the German Hypothesis have also argued that Agatha's marriage is geographically hard to explain. The problem is that an Anglo-Saxon prince, apparently residing in Russia or Hungary, is proposed to have been married to a relative of the emperor, but there is no obvious political context for such a marriage. Vajay and Ronay claim that Eadweard and Agatha were married in 1043 or early 1044 as a result of a triple alliance between England, the Empire, and Russia in 1043 [Vajay (1962), 72-4; Ronay (1989), 118]. However, no real evidence is given for this statement. It is simply a conjecture to explain how an English prince and a relative of the emperor could have been married in Russia. Here, Ronay's attempt to use the Laws of Edward the Confessor on this point to support the German Hypothesis strains credulity. The lack of a convenient explanation how a German prince was married to an Anglo-Saxon exile in Russia or Hungary is a weakness of the German Hypothesis. However, the sources are scanty, and it is not wise to assume that every marriage can be placed in its correct political and geographical context by the surviving evidence. Nevertheless, this does tilt the argument somewhat in favor of scenarios like the Russian Hypothesis in which the marriage is more conveniently explained as a "local" event.

Onomastically, the name Agatha poses a significant difficulty for the German Hypothesis, as the name Agatha is unknown in Liudolf's family. Here, it should be noted that the ancestry of Liudolf's wife Gertrude has not been convincingly demonstrated, so that the name Agatha could have come from Gertrude's uncertain ancestry, or from Liudolf's paternal ancestry, also largely uncertain. In the latter direction, Donald Jackman, who accepts the German Hypothesis as proven, would explain Agatha's name on the basis of a conjectured distant Byzantine ancestry of Liudolf's father Bruno [Jackman (2000), 40-1, 56]. However, this supposed explanation involves too many conjectured links (and too distant a descent) to be convincing. On the whole, the onomastic argument has been overplayed, but it still somewhat weakens the case for the German Hypothesis. See Appendix 1 for a more detailed discussion on onomastics.

If the German Hypothesis were valid, then the empress Matilda would be related to her first husband, the emperor Heinrich V. Heinrich V was the son of Heinrich IV, son of Heinrich III, son of the empress Gisela by her third marriage. On the other hand, Matilda was the daughter of Matilda/Eadgyth of Scotland, daughter of St. Margaret, daughter of Agatha, who would be daughter of Liudolf (assuming the German Hypothesis), son of Gisela by her first marriage. This possible consanguinity, apparently first pointed out by Andrew MacEwen [Faris-Richardson (1998), 235 n. 29], would make Matilda and Heinrich V second cousins twice removed (degree 3:5). While technically within a prohibited degree, it may have gone unnoticed. Marriages of degree 5 often "slipped through the system" and it is therefore not possible to rule out the German Hypothesis on this basis.

Not all of the arguments against the German Hypothesis have been fair. René Jetté, in criticizing this theory, wrote: "In order to satisfy the assertion of the two oldest chroniclers (table 2, extracts 1,2), nieces of an emperor Henry had to be invented." [Jetté (1996), 421] If this were a valid argument, then it would also be an argument against all of the other hypotheses which have been advanced, for in each case Agatha is being identified as an otherwise unknown daughter of some individual. In fact, some primary sources claim that Agatha was a niece of an emperor Heinrich, so it is unfair to suggest that those who use such a source (whether they be right or wrong in doing so) are "inventing" such a niece. As another example, in his criticism of the "German Hypothesis, Norman Ingham wrote the following: "Of more immediate concern, in my view, is the fact that Agatha's grandchildren appear not to have heard about her supposed imperial connections. It would seem nearly impossible that no word of a German tie or a relationship with a Holy Roman emperor reached them. They apparently were aware only that she was related somehow to a king of Hungary." [Ingham (1998b), 261] This statement seems unreasonable, and rather overzealous, for we have no such information about what the grandchildren did or did not know about their grandmother's ancestry. What we do have is a previous series of arguments from the author, by his own admission "very speculative" [ibid., 244], in which it is concluded that "(t)he grandchildren, as far as we can tell, did not subscribe to the imperial idea, no doubt because they had never heard it from their mother." [ibid., 243-4]. Turning speculation into "fact" is not a valid line of argument.

The German Hypothesis is the most natural deduction based on one set of the primary sources. Other primary sources contradict the German Hypothesis, which largely stands or falls on the reasonable, but far from conclusive, argument that the apparently earliest sources which give a relationship to emperor Heinrich are more reliable than those claiming a Hungarian connection. The geographical and onomastic evidence does not fit well, but that is far from decisive. In short, the German Hypothesis in neither so strong as its supporters would claim nor so weak as its critics would have us believe. In my opinion, it is a weak candidate which has the dubious distinction of being slightly more likely than its "strongest" (i.e., least weak) competitor (the Russian Hypothesis).

The Russian Hypothesis

In 1996, when René Jetté published his theory that Agatha was a daughter of Iaroslav I of Kiev [Jetté (1996)], the hypothesis was widely thought to be new. However, Pavsic cited an earlier example [Pavsic (2001), 82 n. 67] and a search of Google Books gives several examples that the theory that Agatha was a daughter of Iaroslav had been around since at least the 1800's [see, e.g., Nob. Univ. France 19: 51 (1840); Miscellanea Genealogica et Heraldica NS 2 (1877): 58; Felch (1894)], although I am unaware of an account earlier than Jetté's which actually sets out the main evidence. In 1998, the New England Historical and Genealogical Register published two articles on Agatha, one highly critical of Jetté's work [Faris-Richardson (1998)] and one highly supportive [Ingham (1998a)]. In the same year, Ingham published a very detailed article on the evidence for the Russian Hypothesis [Ingham (1998b)]. In 2000, an article by Janko Pavsic supported the Russian Hypothesis, concentrating on onomastic arguments [Pavsic (2000); English translation in Pavsic (2001)]. The Russian Hypothesis was also supported in two articles by William Humphreys in 2003 and 2004, although support in the second article wavered to the extent that an alternate scenario (the Byzantine Hypothesis) was proposed [Humphreys (2003, 2004)].

The two main primary sources which have been advanced as evidence for the Russian Hypothesis are William of Malmesbury and the Laws of Edward the Confessor. It is a late addition to the latter source which gives the most direct statement in favor of the Russian Hypothesis. The original version, written probably in the 1130's, states that Eadweard went to Russia, where he was received by king Malesclodus and married to a woman of noble descent ["... ad terram Rugorum, quam nos uocamus Russeiam. Quem rex ipsius terre, Malesclodus nomine, ut audiuit et intellexit, quis et unde esset, honeste retinuit eum. Et ipse Ćdwardus accepit ibi uxorem ex nobili genere, ..." Laws Edw. Conf., c. 35-35.1 (p. 664), see above]. As discussed in Appendix 6, Malesclodus was probably Iaroslav I. Then, an addition made around the year 1200 states that St. Margaret was descended through her mother from the kings of Russia ["ex parte uero matris ex genere et sanguine regum Rugorum; ..." Laws Edw. Conf., c. 35.1*** (p. 664)]. This does not explicitly call Margaret's mother a daughter of Malesclodus, but that would seem to be the natural interpretation. One of the principle pieces of evidence for the Russian Hypothesisis the Gesta Regum of William of Malmesbury, which states that Agatha was a sister of the queen of Hungary. Since one of the queens of Hungary during that period was Anastasia, daughter of Iaroslav I and wife of king Andrew I of Hungary, this supports the Russian Hypothesis.

The testimony of Geoffrey Gaimar has also been used by proponents of the Russian Hypothesis. In mentioning the arrival of the exiled princes in Hungary, Geoffrey gives "Gardimbre" as the name of the city where the king and queen of Hungary resided [Gaimar 4586-8]. Influenced by the fact that a few lines earlier Geoffrey had mentioned passing by Russia on the way to Hungary, René Jetté identifies Gardimbre with a Russian stronghold of Gardorika on Lake Logoda [Jetté (1996), 419]. Norman Ingham states that Gardimbre is probably an Old French rendition of Old Scandinavian gárđr ("fortified town") or less likely gárđaríki ("realm of towns"), and that Gárđr was sometimes applied as a place-name in Old Scandinavian sources to Novgorod or Kiev [Ingham (1998), 239]. Ingham then goes on to suggest that Geoffrey was mistaking Russia for Hungary, saying: "It would seem to follow, then, that his king and queen of Hungary may in actuality represent the grand prince and princess of Kiev and that Gardimbre ought to be Kiev itself." [Ingham (1998b), 239-40]

The proposed parentage of Agatha would fit well geographically and politically with Eadweard's probable exile in Russia. This is a definite point in favor of the Russian Hypothesis as compared with the German Hypothesis. It has been pointed out by Norman Ingham that an eleventh century fresco appears to show that Iaroslav had five daughters [Ingham (1998a)]. If accepted as evidence, this would at least show that Iaroslav had additional daughters above and beyond the ones who had already been identified. As discussed in Appendix 1, it has also been argued that the onomastic evidence favors the Russian Hypothesis.

Critics of the Russian Hypothesis have advanced the following objections:

As is discussed in the following paragraphs, some of these objections are valid and some are not.

In the Laws of Edward the Confessor, the strongest statement about Agatha's origin that she was descended from Russian kings comes only from an interpolation which can be dated about 1200. Jetté misstated the evidence when he claimed that the version of the Laws inserted in Roger of Hoveden calls Agatha a Russian princess [Jetté (1996), 420], and he was severely criticized for this by Faris and Richardson [Faris-Richardson (1998), 225-6]. Indeed, Jetté is wrong on two counts here, because "descended from Russian Kings" is not the same thing as "Russian princess" and because the statement does not appear in Roger of Hoveden but in the later interpolation. On the other hand, Faris and Richardson are wrong when they then rely on the following misleading statements of Ronay: "... the glossarist's description of Agatha as a lady of royal blood related to the ruler of Russia. From the phrasing, however, it is clear that the Englist glossarist had no intention of presenting Agatha as a daughter of Yaroslav the Great. He was simply restating ... that Agatha was of royal blood and had married a royal relation of the ruler of Russia in Russia." [Ronay (1989), 117-8] The words "lady of royal blood related to the ruler of Russia" are also not an accurate description of the words in the interpolation, and Ronay is relying here on the probably false claim that Eadweard's mother was a sister of Iaroslav's wife [see below for more on this]. Commenting on these texts, Norman Ingham acknowledged that neither the original text nor the interpolated text explicitly states that Eadweard married a daughter of the Russian king, but then goes on to say that "(b)oth, nonetheless, imply it, and no other interpretation of either looks plausible." [Ingham (1998b), 255] This is a reasonable interpretation of the interpolated text, which, however, as a late expansion of the original text, has very little authority. On the other hand, the uninterpolated text quite emphatically makes no such implication. It states only that Eadweard's wife was of noble descent ("ex nobili genere"), which in no way implies that she was the daughter of a king or even of royal descent. In isolation, the statement would be consistent with Eadweard's wife being a daughter of the Russian king, but, as noted by John Carmi Parsons, the writer of the uninterpolated text does not give this impression [Parsons (2002), 48]. Since the Russian king had been mentioned in the previous sentence, the writer would probably have identified Agatha as the king's daughter instead of describing her as "of noble descent" if he had really thought that they were father and daughter. Thus, as useful as the Laws of Edward the Confessor are for verifying the exile in Russia and for providing us with Russia as the possible location of Agatha's marriage, the use of this source as direct evidence for the parentage of Agatha is questionable.

This places a higher burden on the information gleaned from William of Malmesbury, the other principle source for the Russian Hypothesis. The Russian Hypothesis requires that William's statement about Agatha's origin be accepted in preference to the contradictory statements contained in several other chroniclers. Ingham and Humphreys have argued in favor of giving preference to William's information [Ingham (1998b), 240-252; Humphreys (2003), 37-42], but the scenarios which they offer for the transmission of the information about Agatha's parentage are extremely speculative. William's information may indeed be the correct version, but as discussed above, there is no compelling reason to accept his account at the expense of the other, contradictory, accounts.

When William states that Agatha was a sister of the queen of Hungary, he does not come to our assistance by identifying that queen ["Filii ejus Edwius et Edwardus, missi ad regem Swevorum ut perimerentur, sed miseratione ejus conservati, Hunorum regem petierunt; ubi, dum benigne aliquo tempore habiti essent, major diem obiit, minor Agatham reginć sororem in matrimonium accepit." Wm. Malmes., Gesta Regum, c. 180 (vol. 1, p. 218)]. As critics of the hypothesis have noted, William's failure to specify which queen of Hungary he was claiming to be Agatha's sister means that it is not clear that he was referring to Anastasia, wife of Andrew I. Thus, the process of elimination (a method criticized by supporters of the Russian Hypothesis when arguing against the German Hypothesis), becomes an important method for identifying the Hungarian queen mentioned by William of Malmesbury. Unfortunately, this point is poorly covered in the papers supporting the Russian Hypothesis, which don't even attempt to list the Hungarian queens during the relevant period, let alone try to rule out the ones other than Anastasia. The problem of other candidates was pointed out by Parsons, who gives a list of the possible queens of Hungary during the period, taken from standard secondary sources [Parsons (2002), 46]. Mladjov briefly considered the two wives (or supposed wives) of Peter Orseolo (king of Hungary, 1038-41, 1044-6) who were mentioned by Parsons, and rejected them as possible sisters of Agatha for "political considerations", stating that "it is virtually impossible that Agatha would have been the sister-in-law of the very monarch her host András I toppled in 1046." [Mladjov (2003), 21, 42-3, 71-2]. While it would undoubtedly be a negative factor, the claim that it is "virtually impossible" is surely an overstatement. Indeed, the slimness of the surviving information makes it difficult, if not impossible, to come up with a complete list of candidates. The queens of Hungary (or wives of kings) during the period are discussed in Appendix 5. It can be seen that there were other queens of Hungary besides Anastasia during the period, some of whom might not even be documented in the records. In fact, one of the other hypotheses which has been proposed (the Polish Hypothesis discussed below) suggests that Agatha was a sister of a different Hungarian queen. Thus, even though the Polish Hypothesis seems less likely than the Russian Hypothesis it is difficult to insist that William of Malmesbury's sister of a queen of Hungary (assuming the information to be accurate) necessarily refers to a daughter of Iaroslav I.

The testimony of Geoffrey Gaimar has been used by the proponents of the Russian Hypothesis in a very questionable way. As noted above, Geoffrey Gaimar is not a reliable source. Much is based on the identification of Geoffrey's Gardimbre, a conjecture apparently based mainly on the coincidence of the first four letters of a nine letter name, which is then used to suggest that Geoffrey's Hungarian king was really the Russian king. In fact, it is simply not appropriate to take an already unreliable source and to emend its statements in some significant way for the purpose of using that altered statement as "evidence" in support of some position, as has been done in this case.

One of the well documented daughters of Iaroslav I is Anna, wife of Henri I, king of France ["... uxorem duxit nomine Annam, filiam Georgii Sclavi regis Rutiorum ..." Ex Chronici Veteris excerpto, s.a. 1047, RHF 11: 159; "Qui post Mahildis reginć humationem, accepit aliam conjugem, videlicet filiam Jurischloht regis Russorum, nomine Annam" Ex Historić Franc. Fragmento, RHF 11: 161; "... ad quemdam regem in finibus Grćcić qui vocabatur Gerisclo, de terra Ruscić, ut daret sibi filiam suam in uxorem" Ex Chron. S. Petri Vivi Senon., RHF 11: 197; "Hic ex Anna filia regis Russić, nomine Buflesdoc" Ex Abbrev. Gest. Franciae Regum, RHF 11: 213; etc.]. John Carmi Parsons has pointed out that if Agatha were a daughter of Iaroslav I of Kiev, then her children, including Margaret, would have been first cousins of king Philippe I of France, and Margaret's daughter-in-law Eadgyth/Matilda, wife of Henry I of England, would have been first cousin once-removed of Philippe and second cousin to Philippe's son Louis VI [Parsons (2002), 43]. Given the close relation between England, Scotland, and France during that time, one would except to see such relationships mentioned in the chronicles if they were in fact true. Yet there is no trace of such relationships in the known sources. While such an argument from silence is not conclusive, it is a strike against the Russian Hypothesis.

Norman Ingham pointed out that an eleventh century fresco "appears to have represented Grand Prince Iaroslav Mudryi" and his wife, he being flanked by as many as five sons and she by five daughters [Ingham (1998b), 231; for pictures, see Ingham (1998a), 216, 222]. If this is an accurate family portrait, it would prove that Iaroslav had more than the three daughters who are known by western sources [most notably Adam of Bremen: "Haroldus a Graecia regressus, filiam regis Ruziae Gerzlef uxorem accepit. Alteram tulit Andreas, rex Ungrorum, de qua genitus est Salomon. Terciam duxit rex Francorum Heinricus, quae peperit ei Philippus." Adam of Bremen, iii, 11, Scholia 63, MGH SS 7: 339]. Ingham admits that the number of sons is not right [Ingham (1998b), 231-2], and suggested that "(s)ymmetry no doubt required five sons to balance five daughters in the composition." [ibid., 232 n. 3] However, as noted by Parsons, if the number of sons is not right, then the number of daughters need not be right [Parsons (2002), 41-2]. Also, since the daughters are not identified anyway, there is no evidence from the fresco that one of them was Agatha, even if we allowed the existence of more daughters (which is likely enough).

Onomastics has played a major role in some of the articles arguing in favor of the Russian Hypothesis [Ingham (1998a); Pavsic (2000, 2001)]. The connection of onomastics to the problem of Agatha's origin is discussed in detail in Appendix 1. Since neither the name of Agatha nor the names of any of her children have been shown to have appeared previously in the families of Iaroslav or his wife, the onomastic argument in favor of the Russian Hypothesis is weak, and has been overplayed, especially by Pavsic. However, Ingham did point out that the names Agatha, Margaret, and Christina all appeared in one of the oldest sources for the availability of saints' names in Russia [Ingham (1998a), 220]. Thus, even though the onomastic evidence is not strong, the Russian Hypothesis does fare a bit better in this regard than its chief competitor, the German Hypothesis. There is an interesting, if questionable, conjecture which would, if true, improve the onomastic evidence for the Russian Hypothesis. In a genealogical table published in 2000, Donald C. Jackman makes Iaroslav I of Kiev a son of Vladimir I by his wife Anna of Byzantium, without offering a source [Jackman (2000), 47]. Although this contradicts the primary evidence, which states that Iaroslav was a son of Vladimir by his wife Rogned, William Humphreys in 2004 suggested that Iaroslav might have been a son of Anna [Humphreys (2004), 284-5; Jackman's table is noted]. In 2008, Jackman gave a detailed argument why Iaroslav might have been a son of Anna [Jackman (2008), 66-75]. The point of onomastic interest is that Anna had an aunt named Agatha. However, the point is very speculative.

In his criticism of the Russian Hypothesis, John Carmi Parsons questions whether Iaroslav would have seen Eadweard ("a landless wanderer") as a viable son-in-law [Parsons (2002), 42]. He points out that in the mid-1040's there was no reason to expect that king Eadweard the Confessor would not have a son. This objection is difficult to judge, because we really know very little about Eadweard's life as an exile, and we don't know the extent to which he was in regular contact with friends in England (if at all). The fact is that there was apparently only one life separating Eadweard the Exile from the English throne, and he would seem to be a fairly attractive candidate for a son-in-law. Thus, it is difficult to accept this as a valid negative argument against the Russian Hypothesis.

As with the German Hypothesis, objections regarding consanguineous marriages have been made against the Russian Hypothesis, but these do not appear to pose any serious problems. One of these objections can be quickly dismissed. Faris and Richardson pointed out that according to information given by Ronay, Eadweard the Exile was a first cousin to the children of Iaroslav and his wife Ingegerd of Sweden [Faris-Richardson (1998), 234]. However, as Faris and Richardson acknowledge, Ronay offers no documentation for his claim that Ingegerd was a half-sister of Eadweard's mother Ealdgyth [Ronay (1984), 45; Ronay (1989), 53 & n. 2 (p. 193)]. Indeed, there is no good reason to accept Ronay's improbable statement [see the page of Ealdgyth]. Of more consequence is a possible consanguinity noted by Andrew Mac Ewen [Faris-Richardson (1998), 234 & n. 28]. Agatha's great-grandson, Henry of Scotland, earl of Northumberland (son of David I, son of Margaret, daughter of Agatha), was married to Iaroslav's great-great-granddaughter Ada de Warenne (daughter of Isabel de Vermandois, daughter of Hugues le Grand, count of Vermandois, son of Anna of Kiev, daughter of Iaroslav). Thus, if Agatha were a daughter of Iaroslav, Henry and Ada would be third cousins (degree 4:4), technically within the forbidden degree. However, such marriages happened often enough that this should not be regarded as a serious objection to the Russian Hypothesis. An example involving fourth cousins once removed (degree 5:6) given by Mladjov would be of no consequence [Mladjov (2003), 66].

The Russian Hypothesis has also been on the receiving end of some unfair criticism. Faris and Richardson write: "The reader, having been informed by Jetté that a previously unknown daughter of Jaroslav has been identified, might expect some explanation why this discovery has been so long delayed, and why she has not shown up in previous work on the princes of Kiev." [Faris-Richardson (1998), 226-7]. For such a statement to be regarded as a reasonable point, it would first be necessary to show that the existing sources are sufficiently comprehensive that it is unlikely that any daughter has been missed by the surviving sources. This is clearly not the case for the daughters of Iaroslav. In fact, the three previously known daughters of Iaroslav are all absent from Russian sources, and appear only in various Western sources.

The Russian Hypothesis depends heavily on the acceptance of the priority of William of Malmesbury as a source for Agatha's origin, and on a specific identity for a vaguely identified Hungarian queen. It does have the advantage of providing a very plausible scenario for the marriage, but on the whole the case has been overstated by its proponents. Comparing the Russian Hypothesis with its main competitor, the German Hypothesis, the direct evidence for the German Hypothesis is significantly better, because the evidence is earlier and the process of elimination involved in the German Hypothesis is relatively straightforward, whereas the search for Hungarian queens produces and alternate candidate and runs into the problem of the thinner Eastern European records, where a complete list cannot be compiled with any confidence. This is mitigated significantly by the geographical and onomastic evidence, which tilt the case to some degree back toward the Russian Hypothesis. While the hypothesis remains somewhat less likely than the German Hypothesis, in my opinion, it is still significantly more likely than all of the other hypotheses.

The Bruno Hypothesis

The hypothesis that Agatha was a daughter of Bruno, bishop of Augsburg from 1007 to 1029, brother of emperor Heinrich II, has been around since at least 1763, when it was argued by György Pray [Pray (1763), 1: 27-8]. In 1779, István Katona supported the Bruno Hypothesis, as did Peter Friedrich Suhm in 1787 [Katona (1779), 1: 260-3; 2: 97-107, not seen by me, cited by Herzog (1939), 1; Suhm (1787), 3: 726]. In 1877, Edward A. Freeman seemed tentatively to reach the same conclusion [Freeman (1870-9), 2: 376, 671-2]. More recently, in 1954, R. L. Grćme Ritchie, after rejecting the Hungarian Hypothesis, concluded that the Bruno Hypothesis was "perfectly tenable" [Ritchie (1954), 392]. The logic of the Bruno Hypothesis is essentially the same as for the German Hypothesis, except starting with a different emperor Heinrich. If one takes the statement of John of Worcester and others that Agatha was a daughter of a germanus of an emperor Heinrich, and then identifies this Heinrich with Heinrich II (d. 1024), one obtains Heinrich's brother bishop Bruno as the obvious candidate.

The most obvious objection to the "Bruno Hypothesis" is that Bruno was a bishop, and that in order for a daughter to be legitimate, she would have had to be born before 1007. Ritchie's claim that "in his day there was no canonical reason for celibacy" is totally unconvincing [Ritchie (1954), 392]. Another very serious problem with the Bruno Hypothesis is that the genealogical appendix of John of Worcester explicitly states that Agatha was a daughter of a germanus of Heinrich III, not Heinrich II. It cannot be directly proven that Agatha was not an illegitimate daughter of Bruno, but it is very improbable.

As a variation of this hypothesis, one could suggest that Agatha was a daughter of Arnold, archbishop of Ravenna from 1014 to 1018, illegitimate half-brother of Heinrich II [Ann. Quedl., s.a. 1014, MGH SS 3: 82; s.a. 1018, p. 84; Thietmar, Chron., vii, 2, MGH SS 3: 837; vii, 49, p. 858]. This is likewise improbable.

The Hungarian Hypothesis

The Hungarian Hypothesis, which makes Agatha a daughter of king Stephen (István) I of Hungary by his wife Gisela, sister of the emperor Heinrich II, has been popular with those who would like to reconcile the contradictory information of the primary sources. The first notice of this theory of which I am aware was in 1778, when Daniel Cornides made this conclusion, called the "Hungarian Hypothesis" [Cornides (1778), 232-9; reference courtesy of Todd Farmerie]. In 1879, Harry Breßlau argued for the Hungarian Hypothesis in his Jahrbücher for the emperor Konrad II [Breßlau (1879-84), 1: 102 n. 1]. The hypothesis was also supported in 1938 by Sándor Fest [Fest (1938)] and in 1952 by G. Andrews Moriarty [Moriarty (1952), 60].

If one wants to reconcile the statements that Agatha was a daughter of the king of Hungary with the statements that she was a relative of an emperor Heinrich, then two kings of Hungary stand out: Stephen I, who married Gisela, sister of Heinrich II, and Salomon, who married Judith, daughter of Heinrich III and sister of Heinrich IV [see Appendix 5]. A quick look at chronology quickly eliminates Salomon as a possible father for Agatha, so that Stephen and Gisela stand out.

The statement of John of Worcester that Agatha was "filiam germani imperatoris Heinrici" is dealt with in two different ways. Breßlau suggested that germanus ("brother") should be emended to germana ("sister"), which, since Heinrich II had only one known sister, would mean Gisela, apparently fitting perfectly with the statement of Geoffrey Gaimar that Agatha was a daughter of the king of Hungary. Fest arrived at the same conclusion by claiming that germanus could mean "brother-in-law" [Fest (1938), 125]. Thus, the Hungarian Hypothesis would seem to reconcile all but one of the main twelfth century sources on Agatha's origin. However, it does this by using an extended meaning of germanus (or an emendation) and by ignoring the testimony of William of Malmesbury.

A number of objections have been raised against the Hungarian Hypothesis, mostly by Vajay and Ritchie:

As noted above, all of the main twelfth century sources which either state or indicate that Agatha was a daughter of the king of Hungary have significant problems. Geoffrey Gaimar is not a reliable source, Orderic Vitalis makes the chronologically impossible statement that Salomon was Agatha's father, Ailred of Rievaulx, after indicating that St. Margaret had royal Hungarian ancestry, goes on to distinguish Agatha from the daughter of the Hungarian king, who is said by Ailred to have married Eadweard's brother. Thus, there is no reason to believe in the reliability of the statements that Agatha was a daughter of a king of Hungary.

The Annales Altahenses Maiores state that Stephen had no surviving sons ["Stephanus bonae memoriae rex, ..., cum filius eius patre superstite esset mortuus, quoniam alium non habuit filium, ..." Annales Altahenses Maiores, s.a. 1041, MGH SS 20: 794]. Some later Hungarian chronicles indicate that Stephen had no surviving children [e.g., "Nam maxime eapropter, ut de suo sanguine dignus nullus esset regni corona sublimari, ..." Simon of Kéza, , Gesta Hungarorum, ii, 24, Chron. Hung., 78; Vajay (1962), 75 n. 3]. It is difficult to accept these statements as completely ruling out daughters. However, Stephen is unlikely to have had any surviving daughters who were politically significant. Any son-in-law of Stephen would have been a potential heir to the Hungarian throne, and the silence of the sources would therefore be difficult to explain unless any such son-in-law remained in the political background for some reason. Even unmarried, a surviving daughter who had not entered religion would have been likely to become a valuable matrimonial pawn in the power struggle that emerged after Stephen's death.

It has also been suggested that any daughter of Stephen would be too old to be the wife of Eadweard or the mother of Eadgar the Ćtheling [Vajay (1962), 71-2]. In fact, the evidence for this is unclear. Vajay states that Stephen became engaged to his wife Gisela in 996 and that all of their children were born between 1001 and 1010, but he cites only secondary sources in Hungarian for these statements [Vajay (1962), 71, 75 n. 2, 9]. In fact, these dates are far from clear. Two German chronicles place the marriage of Stephen and Gisela in 1009 [MGH SS 9: 567, 574]. The birthdate of Gisela is not clear, but her brother Heinrich II was born on 6 May 973 [Hirsch (1862-75), 1: 88], and Gisela could very well have been born in the 980's. Thus, even though it is likely that Stephen's children were older than Eadweard, we cannot rule out the possibility that he had a daughter who was younger than Eadweard. Thus, the chronological objection may not be valid.

Breßlau suggests emending the word germana to germanus. Fest suggests that the word germanus may also mean brother-in-law, a definition which is outside its normal meaning (see Appendix 3). The latter interpretation is especially unlikely in this case, because "daughter of a brother-in-law" is an especially awkward way of describing a niece, and it is very doubtful that a chronicler would use a desription like this. Furthermore, the Hungarian Hypothesis requires that the emperor being mentioned is Heinrich II, whereas the genealogical appendix of John of Worcester shows that it was Heinrich III who was intended. Thus, it is very difficult to argue that the evidence of John of Worcester favors the Hungarian Hypothesis.

As Ritchie noted, king Stephen I of Hungary was canonized in 1083 [Ritchie (1954), 390]. Agatha's daughter Margaret of Scotland died ten years later, in 1093. If she had been the granddaughter of a saint, then that fact would have been well known, and it is extremely improbable that her biographer Turgot would have failed to mention it. Also, the sources which call Agatha a daughter of an unidentified king of Hungary would probably have given Stephen's name if that famous king were her father.

Thus, although the chronological objection is not that convincing, the other objections carry significant weight. Although it has not been ruled out quite so conclusively as some would claim, the Hungarian Hypothesis is still a very improbable solution to the question of Agatha's origins.

It should be noticed that the Hungarian Hypothesis has some further variations. The variation in which Eadweard marries the daughter of a different Hungarian king is discussed in the next section. Another, in which the groom is Eadweard's brother Eadmund, is discussed in Appendix 2. A third variant, in which Agatha marries a step-daughter of king Aba Samuel of Hungary, is part of the Bulgarian Hypothesis, and is discussed under that heading.

A variation of the Hungarian Hypothesis

There are some variations of the Hungarian Hypothesis which have not been proposed by any author to my knowledge, but should at least be briefly discussed. Geoffrey Gaimar states that the wife of "Eadgar" (the name that he uses for Eadweard the Exile) was a daughter of an unidentified Hungarian king. Ailred, in one of his conflicting accounts, states that St. Margaret was descended on her mother's side from the blood of Hungarian kings, but does not specify a line of descent. Thus, we should also try to rule out the possibility that Agatha was the daughter of some Hungarian king other than Stephen.

Jetté rules out this possibility by stating that "a brother-in-law or son-in-law of Peter Orseolo or of Aba Samuel would not have been given lodging by their rival and successor Andrew I." [Jetté (1996), 421] But is that true? Trying to determine political motivation from such scanty evidence is risky business. Stephen I, who, having no surviving sons, favored his sister's son Peter Orseolo as his successor, blinded one of his brother's sons and sent the children of the latter into exile [Annales Altahenses Maiores, s.a. 1041, MGH SS 20: 794]. If the blinded prince was Andrew's father, as seems highly probable, then it is very understandable that Andrew might not want to have any contact with relatives of the pro-German Peter Orseolo. Nevertheless, the old saying "politics makes strange bedfellows" is relevant here. Even though the argument makes a connection less likely, it does not rule it out. Furthermore, there seems to be no reason to believe that Andrew had anything personal against his predecessor Aba Samuel.

Nevertheless, even though Agatha's filiation as a daughter of some other Hungarian king cannot be definitively ruled out, there does not seem to be any good reason to support such a revised Hungarian scenario.

The Cristinus Hypothesis

In his 2002 paper on Agatha, John Carmi Parsons, after discussing the German Hypothesis and the Russian Hypothesis, concluded that the matter was not proven either way [Parsons (2002), 51]. He then followed up by offering "two more theories to indicate just how far we are from the last word on the question." [ibid., 52] First, he asked whether Eadweard the Exile married twice. This possibility, which was not discussed in any detail, would help to explain the different accounts we have of Agatha's parentage, but there is no evidence that Eadweard was married more than once. Still, the question does serve its stated purpose of emphasizing the lack of any definitive solution.

The second theory offered the possibility that Agatha was a daughter of a certain count Cristinus by his wife Oda, daughter of count Bernard of Haldensleben by a daughter of Vladimir I of Kiev. Parsons noted that this descent would give Agatha the Russian royal ancestry suggested by the late interpolation to the Laws of Edward the Confessor, and that the name of Cristinus would explain the name of Agatha's daughter Christina. Also, it is suggested that this would make Agatha a descendant in the sixth generation of emperor Otto I, thus explaining the reference to Agatha as an emperor's kinswoman.

In my opinion, this theory has little to recemmend it. The supposed Russian link, going back only to a great-grandparent, is rather distant, and the onomastic link is tenuous. The supposed descent from Otto I is based in part on links which are themselves very controversial, and even if true, it is unlikely that someone whose only relationship to the emperor was six generations back would be considered a kinswoman of the emperor.

The Bulgarian Hypothesis

The "Bulgarian Hypothesis", due to Ian Mladjov, first appeared in an article published in 2003 [Mladjov (2003)]. Mladjov proposed that Agatha was a daughter of Gavril Radomir (d. 1015), emperor of Bulgaria, a maternal granddaughter of king Géza I of Hungary (d. 997), and a stepdaughter of Aba Samuel (d. 1044), another king of Hungary. This scenario would also give Agatha a paternal grandmother of the same name.

Mladjov's main primary source is a passage in a twelfth century recesion of the Compendium Historiarum of Ioannes Skylitzes by bishop Mikhael of Devol. (Here, I am handicapped by the poor publishing standards mentioned above, in which apparently accented characters are replaced by spaces in the published version of Mladjov's article. I have replaced these spaces with what I think might be the correct characters, in unaccented italics. This affects only proper names, and any errors in the italicized characters are my own.) The English translation (by Mladjov?) of the passage reads:

"His son Gavril, who was also called Radomir, succeeded to the rule of the Bulgarians. He excelled his father in prowess and strength but lagged far behind him in wisdom and intelligence. He was the son of Samuil by Agatha, the daughter of Ioannes Khrysalios, the governor of Durazzo. He began reigning on the 15th of October, in the 13th Indiction. But he did not complete even one year and was killed, when he was going hunting, by Aaron's son Ivan, also called Vladislav, whom he had saved from death when he had been about to perish. Radomir, who was married to the daughter of the king of Hungary, began loathing her for reasons unknown to me and sent her away, when she was pregnant by him. And he married Eirene, a beautiful captive from Larissa." [Mladjov (2003), 50, citing Skylitzes, with the passage quoted in the original language (which I do not recognize, but might be Greek transliterated into Latin characters)]

Mladjov's proposal is that Agatha was the child with whom Gavril Radomir's Hungarian wife was pregnant when he sent her away [Mladjov (2003), 55-6]. The time that the bride was sent away is unclear, except that it was before the death of Gavril Radomir's father Samuil, emperor of Bulgaria, who died in 1014, according to another passage in Mikhael's recension of Skulitzes [Mladjov (2003), 53-4]. Mladjov places this at about the latest possible date, concluding that Gavril Radomir divorced his Hungarian wife and married Eirene in early 1014 [ibid., 54]. Mladjov suggests, reasonably, that the rejected wife would have returned to her Hungarian homeland to have her child. Chronology indicates that this wife's father was probably Géza I (d. 997) and that she was a sister of István (Stephen) I. Mladjov suggests that this princess was then married to Aba Samuel, the later king of Hungary who is called a sororius of Stephen in a thirteenth century Hungarian source, and is therefore believed to have married a sister of Stephen. Thus, if all of this conjecture is true, Agatha would have been a step-daughter of Aba Samuel, fitting reasonably well with those statements which make her a daughter of a Hungarian king. If this is all accepted, then onomastics would be the icing on the cake, for as has been noted, Gavril Radomir's mother was named Agatha. Mladjov even claims that the theory agrees with the statements in the Anglo-Saxon Chronicle that Agatha was a relative of the emperor.

This last claim can be quickly and thoroughly dismissed. The closest connection between a child of Gavril Radomir's Hungarian wife and an emperor would be that the child's (probable) maternal uncle was married to a sister of the emperor. This is clearly not a blood relationship and it is clear that at this point the author was trying to stretch his arguments too far. Even though the Bulgarian Hypothesis is attractive in some ways, there are two serious problems beyond the lack of direct evidence which keep it from being a serious contender. One problem is the identification of Gavril Radomir's rejected wife as the wife of Aba Samuel. This is nothing but conjecture, and the hypothesis is seriously weakened if the identification is not correct.

The other difficulty is the very serious problem of chronology. Gavril Radomir is said to have had at least five children by his second wife Eirene, whom he married after the rejection of his Hungarian wife [Mladjov (2003), 53, 76]. Since Gavril Radomir died in 1015, his youngest child would be born no later than 1016, and probably earlier. Thus, the last child born to Gavril Radomir's Hungarian wife is likely to have been born earlier than 1005, which would be an improbably early birthdate for Agatha. True, it is possible, as Mladjov notes, to conjecture a date of birth for this child as late as 1014 by various arguments, for example, Mladjov's suggestion that Gavril Radomir's second wife was a longtime mistress by whom he already had several children when he married her [Mladjov (2003), 54]. However, this involves further guesswork, and all that Mladjov succeeds in proving is that the Bulgarian Hypothesis is not impossible. In my opinion, it is far from likely.

The Byzantine Hypothesis

In 2004, William Humphreys proposed as an alternate hypothesis the possibility that Agatha was a sister of Anastasia, wife of Iaroslav's son Vsevelod (d. 1093), and apparently daughter of the Byzantine emperor Constantine IX "Monomachos" [Humphreys (2004), 280]. This is the "Byzantine Hypothesis".

The logic of this hypothesis is not very clearly stated, partly because the author tends to dance around the main points by asking questions rather than just making direct statements. Nevertheless, the two main points put forward, neither of them at all convincing, are the following:

Humphreys indicates that when William of Malmesbury refers to the princes taking refuge with the king of the "Huns", this was probably in Russia and not Hungary. While it is true that the princes were probably in exile in Russia before they were in Hungary, it would still be a significant emendation of William's account to suggest that he was calling Agatha a sister of the Russian queen.

With regard to the statement of John of Worcester that Agatha was "filia germani imperatoris Heinrici" Humphreys points out that one of the meanings of germanus is "true" when used as an adjective, suggesting that the "true" emperor referred to the Byzantine emperor. This suggestion has no reasonable basis whatsoever. Not only does Humphreys fail to offer any evidence that the word germanus was ever used in that way to describe a Byzantine emperor, but the presence of the name Henricus clearly shows that it was not intended to be interpreted that way in this case. Humphreys also suggests that Ailred's above use of the term "Imperator Romanus" might refer to an emperor named Romanus (such as the Byzantine emperor Romanus III, d. 1034) rather than the "Roman Emperor". While that would be a grammatically allowable interpretation of that particular sentence taken in isolation, other statements of Ailred rule this out, showing clearly that he was referring to an emperor named Heinrich.

In fact, even the author stated that he considered this scenario to be less likely than the Russian Hypothesis [Humphreys (2004), 287]. The Byzantine Hypothesis should be firmly rejected.

The Polish Hypothesis

The most recent new theory on the origin of Agatha is the "Polish Hypothesis". Proposed by John P. Ravilious in 2009, the Polish Hypothesis conjectures Agatha to be a daughter of duke Mieszko II of Poland [Ravilious (2009)]. Since Mieszko's wife Richenza was a maternal granddaughter of emperor Otto II by his Byzantine wife Theophano, the Polish Hypothesis would, if true, lead to some interesting ancestry for Agatha.

The author starts by pointing out that king Cnut of England and Denmark was related to the Polish royal family through his mother, a daughter of Mieszko I of Poland and sister of Boleslaw I, and thus had close relations with Poland [Ravilious (2009), 71-2]. This connection is well documented by a statement of Thietmar of Merseburg [Thietmar, Chron., vii, 28, MGH SS 3: 848-9]. Ravilious then revives an argument of Johannes Steenstrup that the rex Suanorum/Suuanorum of John of Worcester and the rex Swevorum of William of Malmesbury is not the king of Sweden, but the king of the Slavs [Steenstrup (1876-82); Ravilious (2009), 72-4]. Thus, he argues, Cnut sent the two princes to a Slavic land, and the obvious choice would be Poland, with which Cnut had close relations. The author then points out that until 1031, Boleslaw I and his son Mieszko II of Poland held Ruthenia ("Red Russia") and thus was "effectively the 'king of Red Russia'." [Ravilious (2009), 74-5] This leads to the key identification of the Malesclodus of the Laws of Edward the Confessor as being not Iaroslav, but Mieszko II of Poland [ibid., 75]. With Agatha conjectured as a daughter of Mieszko II, William of Malmesbury is then called in to support the case, for a daughter of Mieszko II was married to king Béla I of Hungary, and thus in this scenario Agatha would also be a sister of a Hungarian queen. The fact that a daughter of Mieszko II and his queen Richenza would also be a descendant of emperor Otto II is then called into play to explain the references to Agatha as a relative of the emperor.

The Polish Hypothesis is similar to the Russian Hypothesis in that both make important use of the Laws of Edward the Confessor and both invoke the support of William of Malmesbury by making Agatha sister of a queen of Hungary. However, the Polish Hypothesis inherits most of the disadvantages of the Russian Hypothesis, while gaining few of the advantages in return. Relying heavily on the testimony of William of Malmesbury, it is subject to the same process of elimination in trying to eliminate other Hungarian queens as possibilities. Furthermore, the attempt to reinterpret the sources by reading "king of Poland" for rex Suanorum/Swevorum or rex Rugorum is unconvincing. This is especially the case with king Malesclodus. The name may look closer to "Mieszko" than to "Iaroslav" at first glance, but the name Iaroslav is known to have had some strange corruptions, and the title rex Rugorum argues strongly against the identification with Mieszko [see Appendix 6 for a discussion of the identity of Malesclodus]. Mieszko holding some Russian territory for a few years would not change this. Also, making Agatha a great-granddaughter of emperor Otto II does not readily explain why she was called a relative of an emperor Heinrich. As shown by Ravilious, the correctness of the Polish Hypothesis would make her a third cousin once-removed of Heinrich III, but it is unlikely that a relationship that distant would be mentioned in the sources. (Her relationship to Heinrich II would be one generation closer, but he was not emperor at the time Agatha married.) In addition, the statement of John of Worcester calling her the daughter of a germanus of Heinrich would have to be drastically emended in order to fit with the Polish Hypothesis, first by changing the gender of germanus to germana (because germanus would have to refer to Agatha's father), and second by seriously revising the meaning of the word germana (see Appendix 3).

Still, the Polish Hypothesis remains possible. The cases for the German and Russian Hypotheses are significantly better, but the Polish Hypothesis seems more likely than the remaining hypotheses.

Wunder's variation of the German Hypothesis

In 1975, Gerd Wunder proposed a variation of the German Hypothesis in which Agatha had a previous marriage:

Conjectured earlier husband of Agatha (very improbable): Vladimir, d. 4 October 1052, son of Iaroslav I.

As we have seen, one difficulty of the German Hypothesis is that it seems hard to explain the circumstances under which the niece of the emperor Heinrich III would be sent to Hungary or Russia to marry an exiled Anglo-Saxon prince. Gerd Wunder, who supported the hypothesis that Agatha was a daughter of Liudolf, conjectured that she was originally sent east as a wife for Vladimir of Novgorod, eldest son of Iaroslav I, and that she married Eadweard only after the death of Vladimir [Wunder (1975), 84-5]. Wunder's conjecture was later supported by Faris and Richardson [Faris-Richardson (1998), 233-4]. However, both Wunder and Faris-Richardson acknowledged that this conjecture is very speculative. Faris and Richardson also hinted that the name of Agatha could have been given to her at marriage as an Eastern Orthodox name, which, if true, would eliminate the onomastic objection to the German Hypothesis [ibid., 232; see Appendix 1]. In addition to the complete lack of supporting evidence, this theory would not leave much chronological room for the children of Eadweard and Agatha. Although it cannot be strictly ruled out on that basis, the theory seems very improbable. It should be noted here that Wunder's variation is not a necessary feature of the German Hypothesis, and that the improbability of Wunder's scenario should not be used to argue against the German Hypothesis itself, as Mladjov does [Mladjov (2003), 38].


The discussion continues on the Appendices page.


Conclusion

Nature abhors a vacuum. In genealogy, one of the corollaries of this seems to be that genealogists are reluctant to regard a parentage as unknown, and would rather pick from among the choices when several alternatives have been proposed. However, this is not always advisable. In my opinion, it seems that we are forced to conclude that, given the current state of knowledge, it is not possible to assign a specific parentage to Agatha which carries conviction. The German Hypothesis and the Russian Hypothesis are are each possibilities, but each also has significant problems. The other hypotheses are all even less likely, and "none of the above" seems like an attractive alternate hypothesis. Pending further discoveries, Agatha's parentage remains unknown. It may not be a popular conclusion, but it appears to be the right one based on the available evidence.


Bibliography

See the Appendices page


Compiled by Stewart Baldwin

First uploaded 20 June 2010.

Minor revision uploaded 27 June 2010 (added early references, courtesy of Todd Farmerie).

Minor revision uploaded 4 July 2010.

Return to Henry Project home page

Go to Henry Project index page

Go to Henry II ancestor table