Einar Ólafur Sveinsson - Dating the Icelandic sagas

8/13/2019 Einar lafur Sveinsson - Dating the Icelandic sagas

1/139

V I K I N G S O C I E T YFOR NORTHERN RESEARCHT E X T S E R I E S

GeneralEditor:G.Turville-Petre


2/139


3/139

V O L U M E I I ID A T I N G T H E I C E L A N D I C S A G A S

By Einar O l. Sveinsson


4/139

The publ ica t ion of th i s vo lumehas been made possible by a gi f t in memory of

D O R O T H E A C O K EBut thewhimhrelstillare callingAnd thesilver water'sfallingJust whereLanga Riverjoinsthe sea.


5/139

D A T I N G T H EI C E L A N D I C S A G A S

A N E S S A Y I N M E T H O D

B YEINAR OL. SVEINSSON

Honorary Life Memberof the Society

V I K I N G S O C I E T YFO R N O R T H E R N R E S E A R C HU N I V E R S I T Y C O L L E G E L O N D O N


6/139

Made and Printed in Great Britain byWestern Printing Services L td, B ristol1958 Einar Ol. Sveinsson


7/139

G E N E R A L E D I T O R ' S P R E F A C E

I HAVE great pleasure in presenting Professor EinarO l. Sveinsson's Dating the Icelandic Sagas. This isa pioneer work, for no general treatment of the subjecthas been published since modern methods of criticismhave been applied.Readers will already be aware of the profundity andthe humane qualities of Professor Sveinsson's scholarship, and many will have enjoyed his earlier works,among which I may mention his sensitive Age of theSturlungs (trans. Johann S. Hannesson, 1953),his penetrating Studies in the Manuscript Tradition ofNjdlssaga(1953), to say nothing of his edition of Brennu-NjdlsSaga(1954),a monum ent of self-sacrificing devotion.The Viking Society is proud to publish a work by sodistinguished an Icelandic scholar, and expresses deepgratitude to him, since he has written this volume especially for us. G.T.P.


8/139


9/139

A U T H O R ' S P R E F A C E

T HIS book was com pleted in Iceland ic in S ep tem be r1956, and nothing has been added s ince then exceptfor two footnotes. The following books and papers, allof which have some bearing on the subject of this work,have been published s ince i t was f inished: HallvardMageroy , Sertekstproblemet i Ljosvetninga (Avhandl ingerutgi t t av Det Norske Videnskaps-Akademi i Oslo II ,Hist.-fil. klasse 1956, No. 2), Oslo 1957; the sameauthor ' s Studiar i Bandamanna saga (Bibliotheca Arna-magasana X V I I I ) , Copenhagen 1957; Ari C. Bouman,Observations on Syntax and Style of Some IcelandicSagas (Studia Islandica 15), Reykjavik 1956; J6nasKristjansson's edition of Eyfirdinga Sggur (fslenzkFornri t IX) , Reykjavik 1956; J6n Johannesson's paperAldur Grcenlendinga sogu (Nordaela, Afmaeliskvedja tilSigurQar Nordals, Reykjavik 1956); Walter Baetke,ijber die Entstehung der Isldndersagas (Berichte i iber dieVerhandlungen der sachs ischen Akademie der Wissen-schaften zu Leipzig, Philol.-hist. Klasse, Band 102,Heft 5) , Berl in 1956; A nn e H oltsm ark 's In tro du ctio n toThe Legendary Saga of Saint Olav (Corpus Cod. Norv .Med Aev. , IV sen, vol . I I ) , Oslo 1956. In this lastnamed work Professor Holtsmark mentions cer tainpalaeographic details , which suggest that the Midsagaof StOlaf was written as early as about 1200.

To my great regret , two dis t inguished scholars , who


10/139

X D A T I N G T H E S A GA Sare m en tion ed in this book, have died since it was w ritten,viz. Professor Jon Johannesson and BarSi GuSmundsson ,Keeper of the National Archives of Iceland.

I wish to express my warmest thanks to the VikingSociety for the honour they have done me in elect ingme an Honorary Life Member , and for including th isbook in their Series.

I would like also to thank my friend Professor G.Turville-Petre for the translation, so carefully made, ofa book which is in many ways difficult. He has sparedno pains to make the translation as exact and readable aspossible.

I w ou ld like finally to th an k M r Pe ter G . Fo ote, whohas kindly read th e type script a nd the proofs , and M rDavid Thomas, who has given a great deal of helpfuladvice on typography and technical problems.

E .O .S .ReykjavikJune1958


11/139

C O N T E N T SGeneral Ed itor 's Preface viiA uthor's Preface ix

I Earlier Researches iII Technical Te rm s 6

III Manuscripts and Texts i iIV Alterations in the Texts 31V Subjective and Objective Evidence 39

VI References to Saga-writing in Early Texts 41V II Th e Ages of M anuscripts 48

V II I Historical Evidence 50IX Family Sagas and Contemporary History 73X Literary Relations 76

X I Linguistic Evidence 96XII Clerical and Romantic Influences 108

XIII Artistry 115XIV Heroic Sagas and the Decline of Realism 124XV Conclusion 127


12/139

A B B R E V I A T I O N SBisk. Biskupa Sogur gefnar lit af hinufslenzka b6kmentafelagi, I-II, 1858-78.I.F. Islenzk Fornrit, 1933-(in progress).Safn Safn tilsoguIslandsogislenzkrabdkmenta, 1855-(in progress).Quotations from Old Icelandic texts are givenin normalized form.


13/139

C H A P T E R r

E A R L I E R R E S E A R C H E S

tiKE many other medieval works , the Icelandic./Family Sagas^ have come down to us without thenames of their authors ; nor do these sagas tel l us whenthey were wri t ten. As might well be expected , cons idering the great value of the sagas , scholars have at temptedto solve such prob lem s. T h e f irst of th em , that of a ut ho rship, has often made the scholar feel as if he were in thepos i t ion of Tanta lus in the Greek myth: the water andfruits which seemed so close were so elusive when hetr ied to grasp them.

The second quest ion, that of the dates when the sagaswere written , has no t ap pe are d nea rly so difficult toanswer, as may be seen from the unhesi tat ing way inwhich scholars have made their asser t ions about theages of various sagas. But if we regard the problemmore closely, we can say that, although there is plentyof evidence, much of i t is exceedingly treacherous. Inthe present essay I shall attempt to examine this latterproblem rather more closely.

At one t ime, scholars used to regard the IcelandicFamily Sagas as t rue pictures of the events , and they

I T h e t e rm Islendinga SSgur (l i teral ly 'Sagas of Icelanders ') is usedin Icelandic for those sagas which relate the l ives of Icelanders l ivingin the Age of Set t lement and the Saga Age (or Viking Age) . Al thoughit is no t al toge ther sui tab le, th e t er m ' Fa m ily S a g a s ' is used -for suchsagas in this book, s ince m an y En gl ish-speak ing people kn ow the munde r t ha t na me . The t e rm Fornaldar SSgur is ren de red by ' H ero icSagas ' , and Riddara Sogur by 'R om a n t i c S a gas ' o r 'Ro m a nc e s ' .


14/139

2 D A T I N G T H E S A G A Strou bled less about the age when they were written; theyasked rather when the events related in the sagas tookplace, and quoted the sagas as if they were contemporarysources. Examples of this approach may still be found,but on the whole it can be said to belong to a past age.First of all, scholars began to notice, as already AmiMagniissonhad done, that many sagas showed in themselves that they were written long after the events relatedin them took place. Gradually this became clearer; allFamily Sagas were written long after their stories tookplace, but some of them appear to be older than others.If sagas are to be used as sources of history, it is obviously a matter of great consequence how long a periodelapsed between the events related and the time whenthe sagas were written.

One of the first attempts to decide the age when thesagas were written was made by P. E. Miiller in hisSagabibliothek (I, 1817), but, as Snorri says in theHdttatal,first attem pts generally leave room for improvement. The dates to which Miiller assigned various sagasdiffer greatly from those to which later scholars haveassigned them, although some of the sagas which hethought were written very late are still thought to be so.I shall not trace the opinions on the dating of FamilySagas expressed by scholars between the appearance ofthe first volume ofMiiller'sSagabibliothek and the timewhen Finnur J6nsson turned his attention to the problem in his great history of Old Icelandic and Norwegianliterature {Denoldnorske ogoldislandskeLitteraturs His-torie,I I , 1898). I should, however, mention that during


15/139

I EARL IER RESEARCHES 3the interval Konrad Maurer published his UeberdieHoensapdris saga (1871), and K. Lehmann and HansSchnorr von Carolsfeld published their book Die Njdls-sage(1883).Both of these books showed a keener appreciation of the criteria of age than was common in thosedays.In his history of the literature, Finnur j6nsson attempted to establish the ages of all Family Sagas. P. E.Miiller had supposed that the sagas had been written atvarious times, but he did not think that there had beenany pause insaga-writingfrom the beginning to the end .But the most striking point in Finnur Jonsson's datingis that he allows for two groups of sagas, one datingfrom about 1200, and the other from about 1300, andbetween these two dates he seems to suppose there wasa gap. This would certainly be a very strange phenomenon, and the doctrine was bound to awaken suspicionand doubt.In his later works and in the second edition of hishistory of the literature (1920-4), Finnur Jonsson expressed more or less the same opinions about the ages ofFamily Sagas. But about the same time, Bjorn M.Olsenpublished his weighty criticism of the Prologue to Stur-lungaSagain which F innur J6nsson had found the chiefsupport for his peculiar system of dating, as will be explained more fully below. Olsen attempted in an independent way to decide the ages of varioussagas, and hereached important conclusions about the relations of

I See B. M. Clsen, Om Gunnlaugs Saga Ormstu-ngu, 1911;UrnttlendingaSogur, published posthumously inSafn V I, 1937-9.


16/139

4 D A T I N G T H E S AG A Svery many of them with the Landndmabdk,which incourse of time provided valuable evidence of their ages.

I regard the bold attempts made in the seriesIslenzkFornrit(1933ff.) and by the editors of this series in otherwritings as a direct continuation of this work of B. M.Olsen. In my view, there can be no doubt that editors ofthe Fornrit have made great advances in assemblingevidence about these questions.But it was not to be expected that such difficult problems would be settled at one stroke, and I have no doubtthat there may be errors in some of the conclusions setforth in these works. At the same time, it should beobserved that the evidence and methods used there havenever been thoroughly examined, and, I think, it isparticularly this which is needed. It is necessary toexamine more closely the basis upon which any datingmust be founded, i.e. the texts themselves, and closerconsideration is also needed of the nature and value ofthe individual criteria which may be used to establishthe age of a work. That was my opinion when I lecturedon this subject at the congress of northern philologists in

Helsinki in 1950. That lecture was, in fact, the basis ofthe present essay.Since that time. Volume VIII B of the seriesNordiskKultur has been published, and it contains SigurSurNordal's remarkable study of the sagas(Sagalitteraturen).When it is realized that the whole study of this extensivesubject covers no more than ninety-four pages, it can

, I See Redogdrelse for n ionde nordiska filologmotet i Helsingfors ochAbo 1951 , p p . 16-17 .


17/139

I EARLIER RESEARCHES 5hardly be expected that the author would find room todiscuss the methods used in dating Family Sagas, andhence I consider an essay of this kind not altogethersuperfluous. For this reason I have given renewed andcloser attention to all those problems upon which Itouched briefly in my lecture of 1950.


18/139

C H A P T E R I I

T E C H N I C A L T E R M S

BEFORE going further I would like to say somethingabout technical term s. Obscurity in the use of technical ternas has seriously hampered scholarly research,and som etimes it has even been difficult to know what thewriter really means. First of all we must consider theword 'saga'. As everyone who knows Icelandic is aware,this word has a very wide application in that language.All the same, in works written in Icelandic during thelast twenty-five years, I think the usage has never beenso obscure as to cause confusion, and this is because of aclear appreciation of the differences. In this essay I shalluse the word 'saga' as it is used, for example, in the introductions to the seriesIslenzk Fornrit.When I speak of sagas in the following pages, I shallalways mean written works, such as EgilsSaga, NjdlsSaga, HeiSarvigaSaga, or Hrafnkels Saga, and I shalluse the word exclusively for such written works. It isimpracticable to use the same word for the sources uponwhich sagas may have been based. It is also impossibleto start out on the assumption, once and for all, that thesources of sagas were of this kind or of that, for such aproblem has to be considered separately in every case.If we suppose that the sources of sagas were oral, thesesources may be called 'tradition'. I have chosen thisword while fully realizing that no more must be saidabout the form of the oral sources than is known. They


19/139

II TECH NICA L TERMS 7may have been anecdotes, shorter or longer tales, or acomplex of short tales. If investigation proves that asaga is based upon a single oral source, a complete oralstory, this will have to be especially emphasized.It may seem strange, but sometimes it appears asthough people overlook the fact that sagas, as we designate them, the written works, really exist, while the oralsources upon which they are supposed to be based, donot exist. We must not close our eyes to the danger,when we feel certain that one incident or another isderived from an oral tradition, that this may be onlyan illusion; this incident may never have existed inoral tradition about the same hero. Scholars shouldnever lose sight of the difference between that whichexists and that which does notthe certain and the uncertain.All who have read works of recent years on the Fam ilySagas must have come across the expressions 'free-prosetheory' and 'book-prose theory'. I shall not expatiate onthese, but I have always considered these terms questionable, since the free-prose theory does not in factallow for a free oral trad ition , but rather for one which ismore or less fixed; though this is of little importance.The chief difference between the two theories is thatthe book-prose theory is not, in the first place, a theory,not in the first place a doctrine, but rather an attemptto follow the tracks from the known to the unknownwithout prejudice, to pass with the help of experienceand probability from one point to the other. On theother hand the free-prose theory, at least in its German


20/139

8 DA TIN G THE SAGASform, is primarily aLehre,a doctrine, which is set forthfully fashioned, and the origin of the Family Sagas isexplained in accordance with it.The advocates of the free-prose theory distinguishsome of the larger sagas, and say that they are moreliterary than the others; they allow for exceptions of onekind or another. With such reservations their doctrinemay be described in this way: the sources of the sagas(i.e. of the preserved, written sagas) were complete oralstories of fixed form, fixed both in matter and in style.In general it is supposed that these complete oral storiescorresponded exactly with the written sagas. It is supposed that they were commonly written down word forword, and it is stated plainly that the differences betweena written saga and its oral source are no greater than thediflferences which two story-tellers would make in tellingthe same tale; the alterations were no greater than anystory-teller would permit himself. These oral stories,from which the written sagas were taken, are believed tohave been learnt by heart by one story-teller from another, and finally, as already said, they were writtendown more or less word for word as one of the storytellers had told the m .As may readily be appreciated, we are here faced witha series of preconceived ideas, I might almost say postulates, none of which follows from the other. The firstpostulate is that when we have a saga its source is a

I K. Liestol (in The O rigin of theIcelandic Family Sagas,1930)was less rigid, and had learnt much more from experience. He hadwide experience of the development of oral tales.


21/139

I I T E C H N I C A L T E R M S 9single, complete oral s tory, although it is obvious thatthere are many other possibi l i t ies , and the scholar ' s taskis therefore m ore com plica ted and difficult . T h e se con dpos tulate is th at every oral s tory wh ich is sup po sed tohave existed was fixed in style and learnt word for wordby heart . But even if we start from the first postulate(which I consider wrong) i t does not follow that thesecond is r ight. I t is well known that there are long taleswhose substance is preserved oral ly, al though they arenot learnt w ord for wo rd. T h is , to speak general ly, is th ecase with the popular Mdrchen. I may add that I cons ider i t quest ionable whether prose texts were ever preserved orally word for word unless there was a distinctclass wh ich lived by reci t ing th em , learning th e m system atically for years together, as the Irishfilid were said todo. There was no such class as this in Iceland.

We come now to the third postulate . I f we suppose forthe moment that the first and second are right, i t doesnot by any m eans follow tha t th e third is r ight . T h e th irdpos tulate is th at th e scribe followed th e oral tale w ord forword. But why should he do so? If I may use a ratherparadox ical expression , I may say th at this kind of wr itingis against the laws of nature. What I mean is this: forcentury af ter century let tered men have been wri t ingdown stories which they learned from oral s tory-tellers ,but what writer was ever so oppressed with a sense of

I This , of course , does not apply to f ixed phrases and suchl ike ,which are preserved in any kind of oral tale . There are also types ofrime-l ike Mdrchen conta ining a lot of repet i t ion, which may be learntmore or less by heart , but in style these are nearly the anti thesis ofFamily Sagas.


22/139

lO DA TIN G THE SAGASinferiority that he did not think himself better able tofind the right words than an illiterate story-teller? It isonly in the last decades tha t scholars have begun to w ritedovm stories word for word, and before they can do itthey have to undergo a strict training. It is such an unnatural thing todo.This is not th e place to discuss the free-prose doctrinefurther, although I shall perhaps do so in another context. Indeed, there is no reason to do so here, for thisessay is not about the sources of the written FamilySagas, but about these sagas themselves. Only in veryrare instances will the free-prose doctrine come in ourway, but yet often enough to make it clear that it shouldbe mentioned and explained briefly. It was appropriateto describe it at the same tim e as I explained what precisemeaning I would give to the word 'sagas' in the followingpages.

I I am concerned with oral prose-stories on profane subjects. Irealize that rather different conditions may apply to stories aboutreligious subjects, but these have nothing to do with the presentproblem.


23/139

C H A P T E R I I I

M A N U S C R I P T S A N D T E X T S

T HE sagas with which I am dealing in this essay arethe extant, written Family Sagas, and when I speakof their age, I mean the date when they were written.But before discussing this, it is necessary to consider oneother problem, viz. the state in which these sagas arepreserved. The importance of the question whether asaga is well or badly preserved has become ever clearerto me during years of study, and much depends uponwhether it is possible to answer it conclusively. T o speakgenerally, most other kinds of scholarly investigation of asaga must depend on the state ofitspreservation. A goodexample of this can be found in studies ofEgilsSaga. Inhis bookForfattarskapettillEigla(1927),Per W ieselgreninvestigated the style of Egils Saga, and among otherthings, he reached the conclusion that the style of thissaga differed so sharply from the style of Snorri that itwas inconceivable that Snorri was its author. But on thispoint SigurSur Nordal refuted him conclusively, showing that what people call the style of Egils Saga is, inreality, the style ofMgdruvallabdk, in which the saga hasbeen considerably abbreviated, as may be seen by comparison with the fragment designated as6.

Family Sagas have come down to us chiefly in manuscripts of the fourteenth and later centuries, for there arevery few noanuscripts, and these only fragments, whichI gils Sagaml F II , 1933, Introduction, pp.Ixxxii ff.


24/139

12 D A T I N G T H E S AG A Scan be ass igned to the thir teenth century. The oldest arethought to be the f ragments 6 of Egils Saga, D2 ofLaxdoela Saga, and the f irs t hand in the vel lum manuscr ip t S th . perg . 18,4to.i Prob ab ly all of the se fragm entsmay be ass igned approximately to the middle of theth i r teen th cen tury . Next come C of Egils Saga and A.M.162, E , fol. containing texts of Eyrbyggja and Laxdcela,an d th e fragm ent S of Egils Saga, which is in a similarhand . Dat ing f rom about 1300 are the manuscr ip ts Diof Laxdoela Saga an d ju s t five m an usc ripts ofNjdls Saga,viz. Reykjabok (A M 468, 4 to) , Grdskinna (Gl. kgl. sml.2870, 4to), Kdlfalcekjarbok (A M 133, fol.) and the fragm ents 8 and ^. The f ragments of Njdla, C, K, y, 6 areassigned to the first half of the fourteenth century, as arethe fragment y of Egils Saga, and the Hauksbdk whichconta ins Eiriks Saga Rauda an dFdstbrosdraSaga (writtenbefore 1334) . The manuscripts Sth. perg. 7, 4to and /3ofEgils Saga are ass igned to the m iddle of th e fourteen thcentury, as are the Wolfenbii t tel manuscript containingEgils Saga and Eyrbyggja and the text of Gunnlaugs Sagain S th . 18.

The grea t ve l lum manuscr ip t Mgdruvallabdk (A M 132I The f ragments o f Egils Saga a n d Njdls Saga denoted by Greekle t ters are con ta ined in A M 162 fol . A-B. I have fol lowed the opinionsof Finnur Jonsson (see especial ly his edit ions of these sagas) and of K.KMund on the ages of these manuscripts. In the fol lowing pages Ihave t aken cons ide ra t ion of F innur Jonsson ' s op in ions , a l though mychief authori ty is KS,lund's catalogue {Katalog over den Arnamag-neeanske Hdndskriftsamling I-I I , 1889-94) .a O n t he ma nusc r i p t s o f Njdls Saga see m y wo rk Studies in themanuscript tradition of Njdlssaga, 1953. In that work I have fol lowedthe op in ions of F innur J6nsson and of KMund in da t ing the manuscr ipts .


25/139

I l l MSS AND TEXTS 13fol.) is slightly older than these last named, and waswritten at some time between 1316 and 1350. It contains Njdls Saga, Egils Saga, Finnboga Saga, Bandamanna Saga, Kormaks Saga, Viga-Gldms Saga, Drop-laugarsona Saga, Qlkofra Pdttr, Hallfredar Saga, Laxdoela,andFostbroedraSaga.Nearly all of these sagas havebeen edited from the Mgdruvallabdk, and if it did notexist, we should have no more than paltry fragments ofKormaks Saga, Droplaugarsona Saga, andViga-GMmsSaga,whileH allfredar Sagawould be preserved only insections, as is e.g.Bjarnar SagaH itdoelakappa.

I do not consider it necessary to enumerate manuscripts later than these, but it is plain that there weresome manuscripts containing various texts written about1400 and in the fifteenth century, which contained twoor more Family Sagas, sometimes together with Heroicand other later sagas. I shall mention only two of these.Vatnshyrna written for J6n Hakonarson ofViQidalstungaabout or a little before 1400, which GuSbrandur Vig-fvisson rescued from oblivion in his Introduction toBdrdarSaga (i860). Vatnshyrna survives only in fragments and transcripts, but it containedFldamannaSaga,Laxdoela,Hcensa-PdrisSaga,VatnsdcelaSaga, E yrbyggja,Kjalnesinga Saga,Krdka-Refs Saga, Viga-Gliims Saga,HardarSaga, Bdrdar Saga,Pdrdar Saga Hredu,BergbiaPdttr, Kumblbda pdttr, and Draumr Porsteins Sidu-Hallssonar. It may thus be said tha t Vatnshyrna andMgdruvallabdk supplement each other. I shall mention

I See J6n Helgason, Gauks Saga Trandilssonar inHeidersskrifttilG. Indrebo, 1939.


26/139

14 D A T I N G T H E S AG A Sjust one other composite manuscript, Melabdk,whichwas written at the beginning of the fifteenth century andcontained Fldamanna Saga, Eyrbyggja, Vatnsdoela, aswell asLandndmabdkand genealogicallists.Considerablefragments of this manuscript arepreserved.*Many vellum manuscripts have been lost, but the lossis partly made good because, early in the seventeenthcentury, people began to copy the vellum manuscripts,and th us we have com plete texts of many sagas of whichonly small fragments survive on vellum. Som etimes thesetranscripts have to serve instead of the lost vellums. Weshall now consider how some of the sagas were preservedunder these conditions. We must consider how manyindependen t texts of medieval sagas now survive in com plete vellums written in the Middle Ages, or in fragments of them, or else in later copies of such vellums.W hen we have to deal with late copies, it does not matterhow many they are, but rather what evidence they provide of the independent medieval manuscripts fromwhich they are descended. It need hardly be said thatinvestigation of all the late manuscripts is necessary, sotha t we may be sure whether they represent one medievaltext or more. Such investigation has only been done inpart, and I must ask my readers to bear this in mind asthey go through the following pages. I must also relyupon the opinions now current. Maybe, the future willbring to light some texts among the paper manuscriptswhich descend from lost, unknown medieval manu-

I Gu&brandur Vigfusson has given a l i s t of other manuscr ipts ofFamily Sagas in his Prolegomena to Sturlunga Saga (1878 ), 29 .


27/139


28/139

l 6 D A T I N G T H E S A GA Snumber of the manuscripts, and the first to which I shalltu rn is that of their age. Needless to say, it does not takelong to alter a saga if a man sets out to alter it. But ifthis does not arise, and if we are dealing with an ordinaryscribe who had no intention of altering a saga, then wecan say that the longer the time and the greater thenumber of intermediaries between our text and theoriginal, the greater are the chances of corruption.Hence, we are in a much stronger position if we have oldmanuscripts, and if they are both old and numerous, weare not altogether at a loss. On the other hand, thescholar must inevitably wander in a maze of bewilderment if he has only paper m anuscripts, representing onlyone medieval codex, whose text may be uncertain inmany passages.T his brings us to the th ird question, tha t of the qualityof the texts. In deciding this subjective considerationsinevitably play a great part, but it is, nevertheless, veryimportant. In considering quality we have the help ofthe learning which scholars have gained by experience oftextual criticism and from the technique of editing, suchas may be found in manuals on the subject. We are in astronger position if we have more than one text, forby comparison we may sometimes acquire some knowledge of the textual history. At the same time, it is aswell to realize that, while many of the changes made inthe texts are undoubtedly corruptions, they may alsoinclude corrections and improvements, although this isless usual. Sometimes a text has been shortened, andthen it may be that the shortened text is every bit as


29/139

I l l MSS AN D TEXTS 17readable as the o ther, and then we need all our wits aboutus. Even in cases where we have only one text, we canstill acquire some knowledge of its quality. We maynotice meaningless phrases, late expressions, sometimesmistakes in personal names, or else the absence of allthese faults. But it is one thing to detect errors, anotherto correct them, and, in doing this, editors have sometimes trusted too m uch to their own judgem ent, and evento their own caprices.I shall now give a survey of the preservation of theFamily Sagas, beginning with Njdla, which survives inthe greatest num ber of manuscripts, for there are relics ofnineteen to twenty medieval codices containing it. Besides these there are later vellums, chiefly fragments,whose texts are, of course, no better than those of papermanuscripts of the same period. Among the paper m anuscripts, traces of a lost codex, probably but not certainlywritten in the M iddle Ages, may be detected. This codexwas the Gullskinna, and it appears to descend from theReykjabdk,which is still preserved. In this case,Gull-skinna would have no independent value. Other papermanuscripts ofNjdlapreserve texts of certain codices ina more complete form than that which they have now.I should point out that not all the paper manuscripts ofNjdla have yet been examined, any more than have thoseof many other sagas, and I ask readers to bear this inmind. As already stated(p.12supra), there are five manuscripts of Njdla, which are believed to date from about1300,and another five dating from the next half-century.If Njdls Sagawas written about 1280,as seems probab le.


30/139

l 8 DA TIN G THE SAGASthe manuscripts are, in age, very close to the original.They thus have the advantages both of number and ofage.I have discussed the manuscripts and text ofNjdla inStudies in the Manuscript Tradition ofNjdlssaga (1953)and in Skirnir (1952), where I reached the conclusionthat, in the original, the text of this saga was similar tothat preserved in its best manuscripts, Reykjabdk,Grdskinna, and Mgdruvallabdk,each of which is the chiefmanuscript ofoneof the threegroups.But the differencesbetween them are so slight that they all constitute oneversion, not only in subject matter, but also in phraseology. In many passages of Njdls Saga it appears to bepossible to attain the original text by comparing thesemanuscripts.In the number of its manuscripts, Egils Saga comesnext to Njdla, for there are thirteen medieval manuscripts of it, although most of them are, of course, fragments.^ In the editions, Mgdruvallabdk has been madethe basis of the text, and indeed no other course seemspossible. Finnur Jonsson divided the manuscripts ofEgils Sagainto three groups, which he designated by theinitials of their main manuscripts, Mgdruvallabdk,Wolfenbiittel, Ketilsbdk.He pointed out that manuscriptsof the W group were abbreviated, and maintained thattheK text was compiled from th e other two. But F inn urJ6nsson did not notice, or did not state, that 6,whichbelongs to the same class asMgdruvallabdk, has a fuller

1 See FinnurJdnsson'sedition ofEgilsSaga,1886-8,Introduction,Ch. I.


31/139

I l l M S S A N D T E X T S I 9text tha n Mgdruvallabdk. SigurQur Nordal d iscussed th isproblem in his edition of the saga, point ing out thatKetilsbdk somet imes shares words and sentences wi th 6which are not to be found in Mgdruvallabdk, and thesame could, in fact, be said of manuscripts of the Wgroup. The explanat ion which he gives is that al l of themanuscripts have been abbreviated, each in i ts own way,except for dwhich canno t be shown to be abb reviated.Thus one manuscript preserves the original text in onepassage, another in another. But 6,w hich is bel ieved todate from about 1250, is only a few decades younger thanthe original text, and appears to follow it very closely.A new understanding of the history of the text is thusobtained. Although the text of Mgdruvallabdk has no tsuffered great alteration, Qgives a different pic tu re of t h eoriginal, which we could not have obtained, at least notwith any certainty, if i t did not exist.

The manuscr ip ts ofLaxdoela a re numerous . There a reseven medieval manuscripts of i t , or their equivalents ,some of which are, of course, fragments . But besidesthese, there are paper manuscr ip ts , which may or maynot descend from yet another medieval codex. I t shouldbe added that chapters from the Laxdoela Saga have alsobeen incorporated in sagas of Olaf Tryggvason and of

1 I.F. I I , 1933, Introdu ct ion, 6 . Reference to th e pro ble m wasalso made in my Introduction to the facsimile of Mgdruvallabdh(Corpus codicum Islandicorum medii csvi V, 1933), pp. 14-15.2 J6n Helgason has la te ly publ ished a s tudy of some paper manuscripts of Egils Saga (Nordcela, Afmceliskvedja til Sigurdar Nordals,1956, pp. n o ff.) . H e points ou t tha t , in cer ta in m anu scr ip ts ofthe seventeenth century, lost vellums have been followed in somechapters .


32/139

2 0 D A T I N G T H E S AG A SOlaf th e S aint . C om pa rison of all these m anu scriptsmakes i t plain that there can be no quest ion of morethan one version of the saga, and their s imilari ty indicates that no great changes have been made since thesaga was first written. It is particularly fortunate that wehave the fragment D2, which must be very close to theoriginal and was probably written only a few years afteri t . T hi s fragm ent sho w s tha t th e text of Mgdruvallabdkis a reasonably sound representat ive of the original ,al though sentences, or parts of sentences, have undoubted ly been omi t ted here and there . We would probably not be able to rely so fully on the evidence of theZ vel lums if we had not got D2. But on this point theevidence of the other manuscripts of the Z group ismuch the same: here and there they contain sentenceswhich are not in the Y g r o u p .^The manuscr ip t s o f Eyrbyggja Saga have been invest igated to some extent in the edi t ions of Gu9brandurVigfusson (Leipzig, 1864), of Gering (Halle, 1897), andin my edition (Reykjavik, 1935), but these investigationsare far from exhaust ive. There are fragments of fourvel lums and transcripts of a f i f th (Vatnshyrna). Mater ia ldifferences are small , and comparison of the manuscriptsdoes not suggest that the wording has been radical lyal tered. I t is possible that manuscripts of the B and Mgroups are, in some part iculars , closer to the originalthan those of the Vatnshyrna grou p, on wh ich the

I T h e mos t im por tan t s tud y of the manusc r ip t s of Laxdcela is contained in K. KMund's edi t ion, 1889-91 ; see also my edition of thesaga in I.F. V , 1934.


33/139

I l l M S S A N D T E X T S 21editions are based. At least, this possibili ty would beworthconsidering.1

There is no thorough cri t ical edi t ion of Grettis Saga ,since li t t le variant apparatus is provided in the editionsof R. C. Boer (Halle, 1900) and G u6 ni Jonsson (Reykjavik, 1936). In his In t roduc t ion Boer maintained thatsix manuscripts of Grettis Saga had independent va lue ,and most of these are medieval vel lums.^ In a few passages there is considerable difference between thesemanuscripts , but I doubt whether i t would be right totalk of two versions.

Pdrdar Saga Hredu has not been published in acrit ical edition. In his edition (Copenhagen, 1848),Halld6r Fri5riksson fol lowed AM 551 d 4to and itst ranscr ip t AM 139 fol. But beside s this the re are fragments of four other medieval manuscripts in the AMcollection, an d ther e is yet ano the r vel lum in Stoc kho lm .I do not know whether any of the paper manuscr ip tshave independent texts . There are two versions of thissaga, one preserved in the fragments of Vatnshyrna, andthe second in the other manuscripts , whose text seems tovary little.Now we come to sagas whose texts are preserved infour vellums (or their equivalents) or even in less. In

1 Mr Forrest S. Scott writes to me (3/6/1957) abou t the remain sof ye t another old text preserved in two paper manuscr ipts (AM446, 4to and 139, fol . ) . This text belongs to the B-class .2 See also Boer 's paper in Zeitschrift fUr deutsche Philologie X X X I ,1899, pp. 40 ff .3 Boerdoesnot men t ion the vellum A M571410,believed to date f romthe s ixteenth century. I t is dif f icult to understand the interrelat ionshipof the paper manuscr ipts f rom Boer 's ar t icle mentioned above.


34/139

2 2 D A T I N G T H E S AG A Ssuch cases, scholars often tend to be less diffident, toplace greater faith in these few witnesses although, inreal i ty, the fewer the texts the greater the caution andscepticism needed, for i t is even more a question ofchance whether al l the versions of the saga have come tolight, and how faithfully they are preserved.

Bdrdar Saga, Kjalnesinga Saga, Finnboga Saga, andViglundar Saga appear al l to be known in three or fourtexts which can be traced to the Middle Ages, but i t isimpor tan t to no te tha t Bdrdar Saga and Viglundar Sagahave not yet been published in satisfactory crit icaledi t ions.

The case of Viga-Gltims Saga is ra th er different. Itm ay not be the m ost im por tant po int that one m anu scr ip tcontains an interpolat ion or, expressed in another way,that one of the manuscripts contains a sect ion not in theoth ers , bu t w hat w ould we say if only this m an usc ript hadbeen preserved? It is no less remarkable, as G. Turvil le-Petre has shown in his edition (Oxford, 1940), that thetext has been shortened in Mgdruvallabdk, w hich is theoldest manuscript, while relics of an earlier text arepreserv ed in tw o fragmentary m anu scr ip ts .

Then there i s the Gisla Saga, which is preserved inthree texts representing two versions, which differwidely in the early chapters , but afterwards draw closertogether . One text i s preserved only in paper manuscripts , and the others in manuscripts which are bel ievedto date from the fifteenth century.

Bandamanna Saga is preserved in three medievalvel lums, of which the oldest is Mgdruvallabdk, wri t ten


35/139

I l l M S S A N D T E X T S 23in the f irst half of the fourte en th c en tury . T h e N o rwegian scholar, Hallvard Mager0y, after s tudying thethree manuscripts closely, has reached the conclusionthat the text of the vel lum fragment in Jon SigurQsson'scollection is possibly descended from Mgdruvallabdk.This leaves us with two independent medieval texts ofthe saga, each of which preserves a dis t inct version. Thetext of Mgdruvallabdk is the fuller, an d vario us the orie shave been put forward about the relat ionship betweenthe two, but , after thorough invest igat ion, Mageroy haslately given the opinion that the shorter text is no otherthan an abbreviated form of the same original text asthat from which Mgdruvallabdk is desce nded .^

In rare instances we may find extraneous evidenceabout the texts of Family Sagas. As examples, I maymention the chapters of Laxdoela which are included inthe sagas of Olaf Tr ygg va son an d O laf the Sa int ( inFornmannasoguran d Flateyjarbdk). T h er e are a lso thr eesagas of poets which have been treated in the same way,except that the Kings ' Sagas preserve a much greaterproport ion of their texts than they do of the LaxdoelaSaga. These sagas areFdstbroedra Saga, Ha llfredar Saga ,and Bjarnar Saga Hitdoelakappa.

Fdstbroedra Saga is preserved separately in four independent texts . The oldest is Hauksbdk, dat ing from thebeginning of the fourteenth century , and then comeMgdruvallabdk and pa pe r t ransc ripts of a lost m an us crip t ,formerly preserved in the Royal Lib rary in C op enh ag en .

I See Kulturhistorisk Leksikon for nordisk Middelalder I , 1956,p p . 332-3-


36/139

2 4 D A T I N G T H E S AG A SThere are also the vel lum fragments in AM 75 e fol . ,w hose te xt is closely related to tha t of a pap er m anu scriptin Stockholm. Besides these, a great part of the Fdstbroedra Saga is preserved in Flateyjarbdk and in twoother related texts , where i t is interwoven with the sagaof St Olaf. There are wide differences between themanuscr ip t s o f the Fdstbrosdra Saga, and th e m ostremarkable is that the text of Hauksbdk is m uc h shorterthan the others . This was formerly held to show that thetext of Hauksbdk was the closest to the original, butSigur5ur Nordal and Sven B. F. Jansson have recent lyargued that the text of the Hauksbdk has been ab bre viated.^

The text of Hallfredar Saga is also preserved in ashorter and a longer form. The shorter i s in the Mgdruvallabdk, w he re the saga is pre sen ted as a s ingle whole;the longer version is in the Greatest Saga of OlafTryggvason, where i t i s broken up in to chapters . Thetext of the Flateyjarbdk is com piled from th at in theGreatest Saga of Olaf and a complete text of HallfredarSaga like that in Mgdruvallabdk. Pr ob lem s of the relat ionship between these two vers ions have not beendecided, and i t may be that both of them have beenaltered.^

Final ly , we have the Bjarnar Saga Hitdoelakappa,which is preserved only in f ragments . There were twovellum leaves of i t in the t ime of Arni Magniisson, who

1 See Bjorn K. p6r61fsson's edit ion of Fostbroedra Saga, 1925-7.2 See I.F. I l l , 1938, p. cxxxix, and especially VI, 1943, p p . Ixxff.;also S. B. F. Jansson, Sagorna om Vinland I, 1944, p p . 173ff.3 Cf. I.E. V I I I , 1939, p. Ixxviii.


37/139

i n MSS AND TEXTS 25got hold of them, but in the seventeenth century more ofthis saga survived on vel lum, and i t had been copied.There are, however, two lacunae in the text , al thoughparts of the opening chapters of the saga have beenincorporated in the manuscripts of the Saga of St Olaf.These chapters have been al tered to some extent , butthey give a fair idea of the content of the first chapters,

Together wi th these sagas I may ment ion Vatnsdoela,preserved in two texts t raceable to the Middle Ages,which do not differ excessively [Vatnshyrna and Mela-bdk and their t ranscripts) . Besides these, excerpts fromthe saga are found in versions of the Landndmabdk, inSturlubdk and Hauksbdk on the one hand, and inPdrdarbdk on the other . These excerpts appear to bebased up on a be t ter text of the saga th an tho se prese rved ,but i t does not seem that there can be any question ofseparate versions, and i t is by no means certain that thetexts varied greatly in style.

We come now to those sagas which are preserved onlyin two medieval manuscripts , and in some cases one ofthe manuscripts is no more than a fragment. Our diff i culties are then graver.There are two vel lum manuscr ip ts of Gunnlaugs Saga,one dat ingfrom the middle of the fou rteenth cen tury , andthe other from the fi f teenth century. There are somedifferences between them, al though their texts are, ingeneral , al ike. One or other must have been interpolatedor shortened.

Kormaks Saga is preserved intact in MgdruvallabdkI SeeI.F. I l l , pp . xcv-vii.


38/139

2 6 D A T I N G T H E S AG A Sand there is , besides, a small vel lum fragment of i t .These two manuscr ip ts have much the same text , wi thonly sl ight verbal differences. There is no reason to suppose that these differences were greater in the chaptersother than those covered by the fragment. This is ofcons iderab le imp or tance , because the Mgdruvallabdk textis so concise that scholars have suspected that this sagahas been shortened and spoil t . But i t may be that wehave here an old, clumsy text , which has not beenser iously corrupted.

Fldamanna Saga is preserved in t ranscripts of oneintact medieval manuscript and in fragments ofanother.^There is some difference between the two texts , the mostremarkable being in the genealogies at the end of thesaga, which must certainly have been al tered in one orthe other .Svarfdoela Saga is preserved only in fragments; thereis one large lacuna in the text and many small ones. Themain text i s found in paper manuscr ip ts which may betraced to a lost vel lum. This text cannot be cal led a goodone , bu t the re is a rat he r be t ter one on a vel lum fragm ent.

Eiriks Saga Rauda an d Droplaugarsona Saga are bothpreserved in two texts , one shorter and the other longer.In the case of Eiriks Saga the longer text (dat ing fromthe fifteenth century) appears to be closer to the originalthan the shorter ( in Hauksbdk), as Sven B. F . Jansso nhas shown, and i t was Law m an H au kr who sho rtenedthe text .

1 See F innur Jonsson ' s ed i t ion , 1932.2 Sagorna om Vinland I , 1944.


39/139

I l l M S S A N D T E X T S 27Droplaugarsona Saga is preserved ent ire in Mgdru

vallabdk, and besides this, a small fragment of i t is foundin a manuscript of the fifteenth century, which was readby Kr . Kdlund in 1886. Scholars have not agreed aboutthe textual his tory of this saga. Konra5 Gislason considered that the text of Mgdruvallabdk had been shortened, but Kilund held that this was the original text ,while that of the fragment had been expanded. J6nJ6hannesson has discussed this quest ion in his edi t ionof the saga,i and has concluded that the text of the fragment is the closer to the original, while that of Mgdruvallabdk has been al tered, not only in wording but alsoin subject matter, especially because of omissions. Theproblem is difficult , but I think that all the indicationssuggest that Jon Johannesson is r ight . We must , therefore, suppose that the original text of DroplaugarsonaSaga has been lost except for the small part preservedin the fragm ent.

The complete text of Hoensa-Pdris Saga is preservedin paper manuscripts which must descend from onemedieval vellum, or perhaps from two, and there is alsoa fragment dating from the fifteenth century. There areno great differences except that most of the completetexts contain an interpolat ion which, according to theopinions of Gu6brandur Vigfusson and SigurSur Nordal ,cannot have been present in the fragment.^

A fragment of Hardar Saga, which belongs to Vatnshyrna, is preserved, and besides this there is another

1 SeeI.F. X I , 1950,p p .Iviiiff.2 See I.F. I l l , 1938, Int rod uct io n, 4 .


40/139

2 0 D A T I N G T H E S AG A Stext. There are great differences between the two (cf.Pdrdar Saga Hredu, p .21 above).

Ljdsvetninga Saga survives only in two late vellumfragments and in paper manuscr ip ts descending fromone of the fragments in a form less defective than thatwhich i t has now, but there is no complete text of thesaga. In this case, there are two divergent versions, oneof which is expanded with tales (pattir) and , in pa rt ,unl ike the other in phraseology. Here we are faced withthe same ques t ion that arose when we considered Bandamanna Saga: can we rely on these manuscripts to giveevidence of the original text? One of them is certainlyunrel iable as a witness, but are they not both? Is notLjdsvetninga Saga wh at m ight be called a bad ly p reserve dsaga?

Jon Helgason has maintained that in the case ofHrafnkels Saga there are two 'branches ' , one preservedin the vel lum fragment and in t ranscripts of i t , and theo ther in paper manuscr ip t s . There a re some, a l thoughno great differences between them. I t should be ment ioned that not a l l the paper manuscr ip ts have yet beenexamined .Final ly we come to those sagas which are preservedonly in one medieval manuscript or i ts equivalent . TheseareHeidarviga Saga, Vdpnfirdinga Saga, Gull-Pdris Saga,Reykdoela Saga, Porsteins Saga Hvita, Porsteins SagaSidu-Hallssonar, and Valla-Ljdts Saga.

I See Hrafnkels Saga Freysgoda, ed. Jon Helgason, 1950 (reprinted1955)- It should be ment ioned tha t J6n Johannesson bel ieves tha ta l l the pap er m anu scr ipts descen d f rom the codex of which a f ragmentis preserved. SeeI.F. X I , 1950, pp. Ivi-vii ,


41/139


42/139

3 0 D A T I N G T H E S A GA SSkdld-Helga Saga: the substance of this saga is pre

served in rimur.The fol lowing lost sagas are named in the Landndmabdk: Bgdmdds Saga Gerpis ok Grimdlfs, Pdrdar Saga

Gellis, Vebjarnar Saga, Isfirdinga Saga (which containedmaterial s imilar to that of Hdvardar Saga). T h e Svarfdoela Saga an d Porskfirdinga Saga ment ioned in theLandndmabdk are gen erally believed to be older versio nsof the extant Svarfdoela and Gull-Pdris Saga.Landndmabdk also co nta ins exc erpts from the followingsagas : *Fljdtshlidinga Saga, *Hrdars Saga Tungugoda,*Kjalleklinga Saga , *Snceb jarnar Saga Galta, *KrceklingaSaga.'^

Certain sagas are also named in other texts:Esphoelinga Saga: nam ed in Pdrarins Pdttr Ofsa. T h e

material is probably included in MrSr's version ofLandndmabdk.

Gauks Saga Trandilssonar: named in Mgdruvallabdk.Material from i t is used in Njdls Saga.

Porgils Saga Hgllusonar: quo ted in Laxdoela Saga.Alfgeirs Pdttr: quo ted in Hardar Saga.In addit ion to these, scholars have deduced the existence of *Porsteins Saga Kuggasonar and of *Porlaugar

Pdttr.^1 T h e nam es of the se are no t cer tain, but I follow J6n Jo han nes son ,Gerdir Landndmabdkar, 1941.2 See I.E. I l l , p. Ixxxi.3 See I.F. X , p p . Ixiv ff.


43/139

C H A P T E R I V

A L T E R A T I O N S I N T H E T E X T S

W E often com e a cross th e op inio n th at th e te xts ofFamily Sagas were, to so speak, fluid, and thatlittle significance can be attached to points of detail. Notonly is this the opinion usually held by disciples of thefree-prose doctrine today, but such conceptions appeartime and again in the works of Gu9brandur Vigfusson.Even in the works of Finnur Jonsson we may often findstatements in which this is implied.

The survey given in the preceding chapter , a l thoughtoo short to do just ice to the top ic, is en ou gh to sho wthat the problem varies from one saga to another. Insome cases we have li t t le evidence about the history ofthe text and in others a great deal . Sweeping statementsand generalizations do not apply, but each case has to bestudied separately.

When we are able to follow the history of a saga-text,this history varies greatly. If we consider the texts ofLaxdoelaor of Njdls Saga, we come to respect them morehighly the longer we s tudy the m . T o be sure , the BollaPdttr is tacked on to the end of Laxdoela Saga in onegroup of manuscripts , while verses have been interpolated in some of the m an usc ripts ofNjdls Saga, but theother manuscr ip ts show the or ig inal form and, by comparison, I think we can get close to the original text ofboth these sagas. In Bandamanna Saga an d Fdstbroedra


44/139

3 2 D A T I N G T H E S A GA SSaga, on the other hand, we find texts which vary cons iderably . Thus , condi t ions are by no means invar iable;so m etim es w e find tha t a tex t is pre serv ed in fairlyconservat ive form, and sometimes that l ibert ies havebeen taken. Whether i t is possible to obtain any rel iableevidenc e of th e h is tory of a text dep end s on the nu m be r ,qual i t ies , and dates of the manuscripts , as was explainedin closer detail in the last chapter.

If a text has been altered, i t may be that i t has beenexpanded or shortened. For example, the longer text ofGisla Saga has been expanded. Formerly scholars generally suppo sed th at texts had b een expan ded, b ut in recentyea rs it has bee n pro ved t ha t the texts of m any sagas hav ebeen shortened, and more examples of th is wi l l probablybe found. But i t would not be r ight to regard this as ageneral rule without invest igat ion, for each text must bejudged separately according to the evidence avai lable.

In the preceding pages we saw that , in some instances,new versions of sagas had been made. In BandamannaSaga, the changes made were chiefly in phraseology, butin such examples asLjdsvetninga Saga, H ardar Saga, andPdrdar Saga Hredu the material of the s tory had alsobeen al tered. Formerly , scholars were very ready to presuppose that there were many versions of sagas, and i twould be a lmost t rue to say that they regarded the huntfor interp olat ion s as the ir chief du ty. I m ust confessthat I often feel that scholars of the latter decades of thenin ete en th c en tury sh ow a cer ta in levi ty in th is . I t wo uldcertainly be foolish to be al together s tubborn and topretend that such th ings never happened, but caut ion


45/139

IV A LTER A TIO N S IN TH E TEX TS 33and care are needed, and every example has to be s tudiedby itself.

In his studies of the Landndmabdk, Bjorn M . O lsenoften allowed for more than one version of a saga and,on the whole, his views were in accordance with those ofhis contemporar ies , a l though he of ten brought moreevidence in supp ort of the m th an o thers d id . F or exam ple,Olsen supposed that there was an older, lost version ofEiriks Saga Rauda, as well as lostPorskfirdinga, Isfirdinga,and Svarfdoela Sagas, and some such sagas are named inth e Landndmabdk. R. C. Boer a t tempted to show thatthere was an older version of Grettis Saga, a n d SigurdurNordal has revived this theory in a new form.

When there is more than one version of a saga it is , ofcourse, the duty of a scholar to at tempt to show reasonswhy such great changes have been made in i t .It should be noted that reasons for altering a saga oncei t had been wri t ten were no t always equally s trong . L et usconsider material alterations. If the author of a saga hadsucceeded in get t ing al l the material from the bestinformed people, i t might well be that he had includedeverything wi th which the s tory was concerned, andthere was then no good reason to add anything. But i fm uch of the m ateria l had been left un use d, the re m igh tthen be good reason to make addit ions, or a new version.The form of a saga might also vary in excellence. If i twas composed with great ski l l in i ts plot , character-drawing, and style, people of later generat ions couldmore easily be content with i t , and it was then less l ikelythat they would make radical changes . The manuscr ip ts


46/139

34 DA TIN G THE SAGASof Njdla, Egils Saga, and Laxdoela seem to give evidenceof this . On the other hand, i t is easy to understand that

' defec ts in th e form of a saga m igh t incite people toi ch an ge it . I thin k th at scholars hav e not given this p oint

the considerat ion i t deserves.In one passage Finnur Jonsson wrote: ' In i t s or ig inal

form no saga began l ike th is . ' Finnur Jonsson had hisown, decided views on what sagas must have been l iketo begin with, at least those sagas which he believedto be oldthey must have been well composed andclassical. These sagas, as he generally believed, had allbeen taken, independent ly of each other , f rom t radi t ion.He bel ieved, moreover, that changes from the originalform were most often damaging. In holding such bel iefs ,the scholar assumes what he has to prove, for i t isobviously just as l ikely that the original work left roomfor improvement and, in some cases , i t was improvedafterwards. As for the relat ions between one saga andanother, the researches of recent decades have shownplainly that saga-wri t ing developed as an unbroken evolut ionary process. In al l probabil i ty, some sagas werepreceded by some kind of his torical summaries inwri t ing. For obvious reasons, most of these are now lost ,but who can say whether some of the oldest sagas werenot also more primit ive in form than later ones? In thepreceding pages I mentioned many lost sagas. Is i t notl ikely that many of them were lost s imply because theywere less highly developed than extant ones?

We may suppose that a new book, or a new version ofa saga, comes into being by the addit ion of s tories


47/139

IV A L T E R A T I O N S I N T H E T E X T S 3 5(pcettir) or by interp ola tion i nto an old er tex t, or elsethat a completely new work is made from the old one,with a new plot and in a new form. I have alreadyremarked that reasons for making a new version mightbe found in the defects of material or form. But anotherreason for alteration might be that there were differentopinions about the people and the events related, otherpoints of view. Bjorn Sigfiisson has suggested that suchdifferences largely accounted for the two versions ofLjdsvetninga Saga, for, in the later version, an at tempthas been made to present Gu5mundr Riki in a ra thermore favourable l ight.

The condit ions rul ing in other cases may be somewhatdifferent. A man might have been described in one sagaor another , but then comes a second author , who takesa different view of the people and events related. He doesnot make a new version of the old saga, but writes a newone, in which the leading characters are different ones.He touches on the material of the older saga and on thecharac ters described in i t, b u t he gives a pi ctu re differentfrom the one given there. I think there were manyinstances of this, and I shall name a few of them.

In Njdla we read a l it tle abou t G u 5 m u n d r Riki .Invest igat ion has shown that the author of Njdla knewLjdsvetninga Saga, but the picture which he gives ofGuSmundr is quite a different one; he describes himaltogether favourably. In his t reatment of I'orkell Hdkrthe author of Njdla shows that he is writing in directopposition to Ljdsvetninga Saga.

In another passage the author of Njdla is certa inly no t


48/139

36 DA TIN G THE SAGAScontradict ing any other book, but he uses the fame ofMQr9r Gigja to enhance the splendour of Gunnarr andNjal l ; for they succeeded where Mgrdr had failed, andgot the bet ter of Hriitr Herjolfsson. It does not seempossible to prove that the author of Njdla knew the lost*Fljdtshlidinga Saga, in which much was told of thewisdom of MgrSrat any rate, he did not have thissaga before him as he wrote, but i t is clear that he knewa b o u t MgrQr's fame and used i t for his own purposes.

It may be seen from Laxdoela Saga that there was awritten saga of I>orgils Hgl luson, and it is natural tosuppose that i t gave a favourable picture of him. Porgi lscomes into the l ives of Snorri Go9i and of Gu9runOsvifrsdottir, and we may suppose that , in spi te of theimpartiali ty of sagas (at least on the surface), the authorof the saga about I>orgils took his part against them.But afterwards another man wrote a saga, in whichGuSriin and Snorr i were among the leading characters ,and he favoured them. He knew the saga about to rg i l s ,and he was careful not to change any of the factualstatements in i t which he bel ieved to be correct , but hedescribed everything in a different l ight. That is how it isin Laxdoela, which contains a direct reference to PorgilsSaga, although this saga is lost , and we have to decidei ts detai ls from probabil i t ies which are, nevertheless ,very s t rong.

I t may also be remembered that the Laxdcela containsa reference to the Njardvikinga Saga. Either this saga isthe same as Gunnars Pdttr Pidrandabana or else GunnarsPdttr is an abbre viate d version of th e Njardvikgina Saga .


49/139

IV ALT ERA TION S IN THE TEXTS 37But we may conclude that the incidents told in i t havebeen used in the Laxdoela Saga, where they have beentouched up in order to embell ish the character ofGu9run O svifrsdo tt i r , and thu s an ent irely differentpicture is given. In all these cases we have to do with anunbroken, organic development. These are, as i t were,debates about people and events; the later author contradicts the earlier.

Not all of the sagas written in the region of theBreiSafjgrdr are prese rved . I have already m en tion edPdrdar Saga Gellisa nd Porgils Saga Hgllusonar. Sigur9urNordal has also concluded that there was a PorsteinsSaga Kuggasonar, and Finnur Jonsson that there was aKjalleklinga Saga, whi le some suppose that there wasan older Eiriks Saga Rauda. We cannot tel l what thesewere like, but other sagas belonging to this district arepreserved, i.e. Heidarviga Saga, Eyrbyggja Saga, Laxdoela Saga, Eiriks Saga Rauda, Groenlendinga Pdttr,which are great artistic unities, the creation of theirauthors . Attempts to show that these sagas have beeninterpolated have not, in my opinion, been successful.The authors of later sagas undoubtedly knew the olderones, and we can often see that they did. Sometimesauthors of the later sagas take account of what is said inthe older ones, e.g. they take care not to repeat i t , or theycontradict the older sagas by tell ing the story differently.The surviving sagas of the Brei6afj(jr9r are, for the mostpart , well constructed.

Sagas of this district , as indeed those of most otherparts of Iceland, appear to have been fixed unities ever


50/139

3 8 D A T I N G T H E S AG A Ssince they were first written. But in EyjafjgrSr therewere a remarkably large number of short tales {pcettir),wri t ten but more or less unat tached. The older vers ionof Ljdsvetninga Saga is very loosely constructed, andmany short tales have been inserted in the later version.I think it correct to say, as some scholars have stated,that the version of Viga-Gltims Saga in Mgdruvallabdkhas bee n interp olate d, an d a furth er tale {pdttr) has beenadded in the Vatnshyrna. T h e rela t ions between theEsphoelinga Saga and the Viga-Gliims Saga are notal together clear, and the Valla-Ljdts Saga is more likea tale {pdttr) th an a com plete saga. Som e scholars havededuced from the re la t ions between Reykdoela Saga andViga-Glums Saga tha t there was an independent Por-laugar Pdttr. It may well be tha t m eth od s of saga-writing in this region were rather different from thosefollowed, for example, in BreiQafjgrSr and BorgarfjgrSr.If so, i t would be wise to tread cautiously; the basisfor dating the sagas of the EyjafjgrSr district may beless secure.


51/139

C H A P T E R V

S U B J E C T I V E A N D O B J E C T I V EE V I D E N C E

IN 12 of his Prolegomena to the Sturlunga Saga ,GuSbrandur Vigfusson discusses the ages of theFamily Sagas. It is as if wisps of evidence pass by, oneafter the other, and he is aware of their weaknesses andof the danger of t rus t ing them. But there i s somethingleft , which does not let him down. He writes:

The best we can do is to look at the tone and character ofa saga, which even in a late adulterated form is never quiteeffaced; even the worst, Svarfdcela, shows marks of antiquity.No furbishing can hide the antique grace of a true saga, suchas Kormak's or Havard's, neither can any affectation ofspurious age make Kjalnesinga orViglund'sSaga look ancient.

Th ere is no doub t that G u9 br an du r Vigfusson had akeen appreciat ion of the Icelandic language and ofIcelandic sagas, but might he not have failed to distinguish between 'ant ique grace ' and simple 'grace '?

The authors of sagas looked back, for they were t ryingto describe an age which had passed. These authorsdiffered in taste and tem pe ram en t, as th e sagas show ; b u tbetween the composit ion of many sagas only a few yearsor decades could have elapsed. How could i t , then, bepossible for anyone to decide their ages simply from'tone and character '? I t is , indeed, certain that few wouldnow accept the assert ion that Hdvardar Saga is ancient .

So much for the green tree, but what of the dry? What


52/139

4 0 D A T I N G T H E S AG A Sare we to say w hen T o m , D ick, and H arry th ink that theycan establish the ages of sagas from their ' tone andcharacter '? I th ink of young men who are put to wri tetheses after studying sagas only for a few years, whenthey have perhaps read few of them, and those quickly.

N o , it is obvious that fortune will not favour endeavours of this kind. Instead, we must search with al l themeans available for concrete, objective evidence, and tryto estimate the value of every shred of i t . In very fewinstances can we speak of absolute certainty; generallyit is rather a question of probabili t ies, whose strengthwe must assess. I do not despise the feeling for ' tone andcharacter ' , but some basis must be found for i t ; i t mustbe turned into logical argument. Above al l , we mustavoid confusing 'characteris t ics ' with 'characteris t ics ofage' .Accordingly, I shall look first for concrete, objectiveevidence, and try to decide the value of i t in every case.Afterwards, I shall come to the less concrete evidence,including the ' tone and character ' , if i t is ever possibleto get a grasp of such things.

My aim is not to argue a case, not to 'prove' but , inthe first place, to find out where we stand, what we know,and what conclusions we may think i t proper to draw.


53/139

C H A P T E R V I

R E F E R E N C E S T O S A G A - W R I T I N GI N E A R L Y T E X T S

IT is natural tha t , in the ir at te m pts to establ ish theages of Family Sagas, scholars have tried to findgeneral principles, according to which the ages of allsagas could be established, or at least the ages of someof them. I t is thus natural to conclude that sagas inwhich the civilization of the Icelandic republic is described in close detail and with verisimilitude were mostlywri t ten during the age of the republic, or not long afterwards, since the succeeding age differed in many ways.This seems to accord with other evidence, but i t doesnot suffice for precise dating.

Scholars have tr ied to date a great proport ion of thesagas by other means; they have looked for passages incertain ancient wri t ings in w hich the au tho rs have spoke nin general terms about the ages of sagas. When theauth or of the First G ram m atica l Tre at ise l is ts th ebranches of l i terature which it was fashionable to writein his day, he mentions law, genealogy, renderings ofrel igious works, and those learned works which Arihad compi led. The Grammatical Treat ise cannot bedated with absolute certainty, but despite all objections,I think i t was probably wri t ten about the middle of thetwelfth century. From the words of the Treat ise i t seemsthat we may conclude that sagas were not commonlywrit ten at that t ime. On the other hand, in the Prologue


54/139

4 2 D A T I N G T H E S A GA Sto the Hungrvaka, w ritt en in th e first decade of thethi r teenth century , ment ion is made of ' that which iswri t ten in the Norse tongue, law, sagas, and familyhistory' .^ To be sure, the author does not say what kindof sagas these were. Kings ' Sagas, Family Sagas, orboth; but i t is certain that , in his t ime, wri t ten sagasmust have been known and fai r ly common.

Snorri , in his Prologue to the separate Saga of StOlaf, also tel ls us something about saga-wri t ing: 'Morethan two hundred and for ty years had elapsed s ince thet ime when Iceland was set t led before people began towri te sagas . That was a long per iod, and there was therisk that oral t radi t ions would have been corrupted . ..'^I t is not al together clear to what period Snorri al ludedin the words 'when Iceland was set t led ' , but even if hemeant the beginning of the set t lement , he could notimply that saga-wri t ing s tar ted before c. 1120. But itmay be that he referred to a period twenty to forty yearslater. Plainly, Snorri was thinking chiefly of the Kings'Sagas , but he speaks in such general terms that i t isnatural to conclude that no sagas, and therefore noFamily Sagas , were wri t ten before .Of all that was said in ancient works about saga-w ri t ing, th e w ords of the so-called ' P ro lo gu e' to Sturlunga Saga have undoubtedly had greates t consequence.Th i s 'P ro logue ' s t ands be tween Sturlu Saga and thePrestssaga Gudmundar Gdda, and it was certainly written

I 'pat, er anorroenu er r i tat , Igg e5a sggur e6a mannfroeSi. 'z 'Pat var meirr en tvau hundruS vetra t6Ifroe8, er Island varbyggt , a&r menn tceki h6r sggur at r i ta , ok var }>at Igng sevi ok vant,a t SQgur hefdi eigi gengizk 1 munni . . . '


55/139

V I R E F E R E N C E S T O S A G A - W R I T I N G 4 3b y t h e m a n w h o p u t t h e Sturlunga c o l l e c t i o n t o g e t h e r ,I>6r9r N a r f a s o n , o r w h o e v e r i t w a s . T h e 'P r o l o g u e ' ^ r e a d sas fo l lows:

Many of the sagas herewillbe contemporaneous, but yetthey cannot all be writ ten atonce: the sagas of BishopI>orIakr th e Sain t an d ofGuQmundr Arason the Goodup to the t ime when he wasordained priest; the Saga ofGuSmundr Dyri begins threeyears after the death of Sturlaand ends after the death ofBishop Brandr , and Gu5-mundr the Good was thenconsecrated Bishop; the Sagaof Hrafn S veinbjarn arson andPorvaldr Snorrason is contemporaneous with that ofGuQmundr the Good, and i tends after the death of BishopBrandr, as Sturla I>6r9arsonrelates in theIslendinga Sggur.Nearly all the sagas whosestories were set in Icelandwere writ ten before BishopBrandr Saemundarson died,but those whose stories tookplace after that were rnostlynot writ ten until the poetStur la I>6r9arson compiled

Marga r sggu r ver9a hersamtiSa, ok ma JJO eigi allarsenn ri ta: saga Porlaks bisk-ups ins helga ok G u 9 m u n d a rin s goSa Arasonar , Jjar til erhann var vfg9r ti l prests; sagaGuSmundar ins dyra hefskJjrimr ve t rum ep t i r and la tStur lu ok lykr J?a, e r Brandrb iskup e r anda9r , en Gu9-mundr inn g69 ivigSr til bisk-ups ; saga Hrafns Sveinbjarn-arsonar ok torvalds S n o r r a -sonar er samtiQa sggu G u d mundar ins goQa, ok lykskhon ep t i r andldt Brands b isk-u p s , sva sem Sturla I>6r9ar-son segir 1 fs len ding asg gum .Flestar allar sggur, )jaer erher hafa ggrzk dIs lan di , varuritaSar, a9r Brandr b iskupSaemundarson andaSisk, enJjasr sggur, er si9an hafaggrzk, varu litt ritaQar, a9rSturla skald I>6r9arson jag9ifyrir Islendingasggur, okhaf9i h a n n j^ar til visendi affr69um m g n n u m , ]?eim erva ru a gndve r9 um dggum

I Sturlunga Saga, ed. K. K lund, 1 9 06 - 1 1 , I , 119-20 .


56/139

4 4 D A T I N G T H E S AG A StheIslendinga Sggur;and for hans, en sum t eptir br^fumthis he relied upon informa- ]?eim, er Jjeir ritu9u, erJ>eimtion supplied by knowledge- varu samtida, er sggurnar eruablemen who had lived in his fra. Marga hluti mdtti hannearly years; some of it he got sjalfr sja,J>d erdhans dggumfrom records written by people ger9usk tilst6rti9enda. . . .who lived at the same time asthose of whom the stories tell.He had also been able to witness many momentous eventswhich took place in his ownday. . . .

When i t says in this text that Sturla wrote the Islendinga Sggur, the work referred to is that commonly knownnow as the Islendinga Saga, which tells of events in theStur lung Age. But many scholars bel ieve that the words'nearly all the sagas . . . {Flestar allar sggur . . . ) ' referto the works which are now cal led Family Sagas. I t wasprobably P. E. Mii l ler who firs t put forward this view,^and he concluded from this passage that the FamilySagas were mostly wri t ten before 1201, the year whenBishop Brandr died. Many later scholars shared thisopinion, and i t was especial ly important that FinnurJ6nsson advocated it in his great history of the l i terature.Many scholars , however, cri t icized this interpretat ion,including N. M. Petersen, Gu9brandur Vigfusson, andKonrad Maurer . That percep t ive scholar Gu9brandurVigfusson fel t sure that there was something quest ionable in the text , and when the second vel lum was examined, i t was found to contain the very sentence which

I Sagabibliothek I, 1817,p. 248.


57/139

V I R E F E R E N C E S T O S A G A - W R I T I N G 4 5Gu9brandur Vigfusson had guessed should be there.Bjorn M. Olsen discussed this problem in three works^and, as it seems to me, it was he who finally solved it.The text quoted fol lows the Krdksfjardarbdk, on e ofthe two main manuscr ip ts of Sturlunga Saga. This t ex tis a little strange, since it gives no chronological limitsfor these sagas, not saying when the events related inthem took place, but only that they were wri t ten beforethe death of Bishop Brandr. There are other diff icul t iesas well . Of the sagas direct ly named in the 'Prologue' , i tis known that Gudmundar Saga Dyra an d PrestssagaGudmundar Gdda must have been wri t ten a considerablet ime after the de ath of Bishop B ran dr, so th at th e ev idence of the Krdksfjardarbdk, even abo ut the sagasnam ed, is w rong . If th e F am ily Sagas we re inc lude d aswell , we should have to admit that some of them musthave been wri t ten after the death of Bishop Brandr, andevidence in support of this will be given later in thepresent essay.

But investigation has shown that in Reykjarfjardarbdk,the o ther vel lum manuscr ip t of Sturlunga, the text ofthe passage is different, and reads:

Nearly all of those sagas Flestar allar sggur, Jjaer erwhose stories were set in Ic e- ggrzk hgf9u i Islandi, a9rland before Bishop Brandr Brandr biskup SasmundarsonSxmundarsondied were writ- anda9isk, vdru rita9ar, enten, but those which took Jjser sggur, er si9an hafaplace afterwards were mostly ggrzk, vdru litt rita9ar, a9r1 Um Sturlungu in Safn I I I , 1897, pp . 193-510; Om den sdkaldte

Sturlungaprolog, 1910; Um Islendingasogur in SafnYl, 1937-9 .2 Sturlunga Saga, edit ion cited above I. 120.


58/139

4 6 D A T I N G T H E S A GA Snot written before the poet Sturla skald I>6r9arson sag8iSturla I>6r8arson com piled fyrir Islendinga Sggur.theIslendingaSggur.If these words were taken to apply to those sagaswhich deal with the 'Saga Age' ( i .e . Family Sagas) , thelat ter reading would only tel l us that they were wri t tenbefore the Islendinga Sggur of Sturla, which he can hardlyhave s tar ted to compi le before the mid-thi r teenth century, or even later. But is i t really true that the authorof this passage was talking about Family Sagas? Is notP . E. Mii l ler ' s in terpreta t ion wrong on th is point?The Prologue is otherwise concerned only with sagasin the Sturlunga collection an d sagas ab ou t later t im es.B jorn M . O lse n show ed tha t th e difficulties disap pea r ifthe text is taken to apply to the so-cal led contemporarysagas. Porgils Saga ok Haflida, Sturlu Saga, and PorldksSaga all end before 1200; Prestssaga Gudmundar ends in1203, but is concerned chiefly with the t ime before thedeath of Bishop Brandr, and the chief events related inth e Gudmundar Saga Dyra also took place before 1203.

Bjorn M. Olsen did not let the matter rest there. Healso showed that the Sturlunga Prologue was compiledfrom one which Sturla had wri t ten himself and hadplaced at the head of his Islendingasggur, and the sentences under d iscuss ion would there have read:

Nearly all of thosesagaswhose stories were set in Iceland before Bishop Brandr Saemundarson died have been written, butthose which have taken place sincehavemostly not beenw ritten.I 'Flestar al lar sggur , )?aer er ggrzk hgf8u a Is landi, a3r Brandrb iskup Sasmundarson andaQisk, e ru ritaSar, en Tpser sggur , er sidanhafa ggrzk, eru litt r ita5ar. '


59/139

V I R E F E R E N C E S T O S A G A - W R I T I N G 4 7Since this was the case, since the sagas of the later

period had been l i t t le wri t ten, Sturla set to work andbegan to compile his book. Although in his IslendingaSaga he mentions some events which took place beforethe death of Bishop Brandr, the detai led account givenin i t begins jus t after the d eath of Bishop B ran dr. W h enwe real ize this , the whole problem becomes clear. Thisalso explains why phrases which were natural as Sturlawrote them have become clumsy in the Sturlunga P r o logue, w ith tw o parallel m ain clauses w ith relative c lausesinserted, while the ending necessary for both of them (thesubsidiarytemporalclause) does notcomeun ti l afterw ards.

The conclusion, therefore, is that all evidence suggeststhat the Sturlunga Prologue has no relevance to theFamily Sagas, and even if it had, it would only tell usthat many of them were wri t ten before the middle ofthe th i r teenth century . Thus , we can make no use ofP . E. Mii l ler 's interpretat ion of this Prologue in establishing the ages of the Family Sagas.

On the other hand, the evidence about saga-wri t ingwhich I quoted ear l ier , f rom the Grammatical Treat ise ,the Saga of St Olaf, and the Hungrvaka, is so un d andreliable as far as it goes. But if we are to learn moreabout the ages of individual sagas, we must look forevidence of other kinds, and to this I shall now turn.^

I T h e man usc r ipt S th. pe rg. 2, 4to contains a l is t of boo ks, wri t tenin a han d said to date f rom abo ut 1300. The list consists of: SkJQldungaSaga, Rumverja Saga, two books , Sturlu Saga, Draum a Saga,Eyrbyggja, Qnundarbrennu Saga, Viga-Glums Saga. See Saga OlafsKonungs bins Helga, ed. O. A. Johnsen and Jon Helgason, 1941,I I , 886. This l is t is interesting, but i t would have greater value if i twere older.


60/139

C H A P T E R V I I

T H E A G E S O F M A N U S C R I P T S

F IRST of all we have to consider the ages of manuscripts . Only very few manuscripts of Family Sagasare o lder than 1300, but we should probably not at tachtoo much significance to this. It is always dangerous todraw has ty conclus ions ex silentio, unless there is otherevidence as well . I t may be mentioned that the profanesagas of contemporary life (chiefly preserved in theSturlunga collection) are not found in m an usc ripts olderthan the fourteenth century. In so far as sagas did notcirculate widely, and existed only in a few copies, i tcould hardly be expected that more than very few copiesshould be preserved, and it is less l ikely that the olderthan the later manuscripts should survive. Since we haveso few very old manuscripts, the ages of sagas are general ly del imited more closely by other evidence. Nevertheless, it is sometimes useful to consider the ages of manuscripts. It is , for example, of great significance that theoldes t manuscr ip ts of Njdls Saga are believed to datefrom about 1300, and that there are f ive of these. Theirlarge number assures us that the saga is not younger thanthis date, for al though the dat ing of manuscripts isgeneral ly somewhat uncertain, i t is unl ikely that theyshould al l be supposed to be later than they real ly are.I t is also surpris ing that so many manuscripts shouldappear a t the same t ime, and that there should be noneolder. On the other hand, we must suppose that some of


61/139

Vn AGES OF MSS 49the manuscripts have been lost , so that the saga mustbe a good ten or even twenty years older than 1300.Finnboga Saga is preserved only inMgdruvallabdk which ,as already said, was wri t ten between 1316 an d 1350,shal l we say 1330-40. There i s nothing to suggest thatthe saga is much older than that , and the evidence of themanuscript is valuable, for i t gives a terminus ante quem.

The manusc r ip t AM 162 D2, containing par t ofLaxdcela Saga, is bel ieved to date from ab ou t th e m idd leof the thir teenth century, and this is also important , forwe might otherwise suppose that the saga was wri t tenrather later than that .

The oldest manuscripts of Egils Saga, Eyrbyggja Saga,and Heidarviga Saga are not wi thout importance, but inthese cases, other evidence shows that the sagas belongto a s t i l l older period than these early manuscripts .T h u s , the age of manuscripts is sometimes of value inestablishing the ages of sagas. But we cannot base conclusions on the fact that a saga exists only in late manuscripts, least of all if the medieval manuscripts of i t arefew. To prove that a saga was wri t ten late we need otherevidence.


62/139

C H A P T E R V I I I

H I S T O R I C A L E V I D E N C EW E com e no w to wh at we m ay cal l 'historical evidence', i .e. a saga mentions people or events ofan age later than that with which it is concerned, or elseit alludes to practices or details of antiquity which givesome evidence of i ts age.

First of all we may mention the closing sentences ofth e Droplaugarsona Saga, which are very famous, andnot without reason. I t says there:

After Ingjaldr was dead, Helga lived at Arnei9arsta5ir, aswell asI>orkell, son of Helga and Grimr. I>orvaldr had a soncalled Ingjaldr, and his son wasI>orvaldr, who told this story.A year after the priest I>angrbrandr came to Iceland, HelgiDroplaugarson waskilled.'

This is the only instance in which i t is said that aFa m ily S aga wa s 'told ' , and in w hich the ' tel ler ' is nam ed .But as soon as the passage is examined a l i t t le moreclosely, i ts faults come to l ight. Mention is made oft o r k e l l , the son of Grimr, but in the next sentence i tsays : 'I'orvaldr had a son. ' I t is obvious that some mistake has been made, and the usual explanat ion is that'I>orkeir is a mistake for 'I>orvaldr'. This is an old correction, for i t is found already in the text of the Fljdts-doelaSaga. We might thus suppose tha t 'I>orvaldr' was

I 'Helga bjo eptir Ingjald l iSinn a Arnei8arstg5um ok Porkell,s on r Jieira Gr ims. porvaldr a t t i son, e r Ingja ldr h^t. Hans sonr h6tPorvaldr , er sag3i sggu Jjessa. Ve t r i si5ar en Pangbrandr p res t r komti l I s lands fe l l Helgi Droplaugarson. '


63/139

V I I I H I S T O R I C A L E V I D E N C E 51the original reading. In that case, the man who told thestory must have l ived about the same t ime as Ari theWise. The faul t might also have arisen by dropping oneor more generat ions out of the genealogy, and this waswhat Bjorn M. Olsen supposed. H e looked for m enliving in the Sturlung Age, who might have belonged tothis family. Jon Johannesson has suggested, on the otherhand , tha t the original auth or of th e Droplaugarsona Sagadrew on a wri t ten account of these brothers , a summarylife, in which I>orvaldr was ci ted, but the compiler ofthe version in Mgdruvallabdk misunders tood th i s , andthought tha t torvaldr was the source of the saga itself.Jon Johannesson m ent ions two smal l points which wo uldbe explained on this supposit ion, but, as i t seems to me,his bril l iant surmise has too li t t le basis, and the explanat ion given by Bjorn M . O lsen, tha t a pa rt of the ge ne alogy has dropped out , is more probable. I can nei theraccept nor reject Bjorn M . O lson 's at tem pts to f inddescendants of this family among the people of theSturlung Age, but i f one or more generat ions have beenomitted from the genealogy, the t ime-l imit given for theage of the saga would not be close, and it would be implied only that i t was written in the l

Documents

Einar Ólafur Sveinsson - Dating the Icelandic sagas