Karl has repeatedly stated the he never heard performances worth mentioning in his early life. "There were a few who were trying, but they were completely out of tune (laglausir) and did not know any melody and not any bragur either." Nevertheless he was "thoroughly trained" in his late thirties when he met his father-in-law, Sumarlii Gumundsson, the only performer he clearly remembers. "But I did not use his bragur however."
A similar attitude appears in Thórur's remarks on kvæaskapur given as an answer to the question whether he learned to kveda from Brynjólfur Björnsson, the great kvæamaur from his boyhood: "No, there was such a variety. Not everyone had the same voice quality, but the bragur should really be the same [i.e., during a performance]. But it is not easy to imitate anyone in this respect -- not directly. There is for instance Snæbjörn from Hergilsey whom many know from hearsay; he did kvea in a somewhat different manner from many others. For example, he begins in this way [Thórur performs No. 23 at the bottom of the register (pitch and seimur near "g") in a dark, hollow, slightly trembling voice]. It was as if they felt their importance in one way or other, and I don't really know the reason why this was not his natural voice quality. I cannot imitate people precisely. I cannot do that but this is not far from it...But I doubt if it is really possible to teach it [kvæa ] to others. As we can see, sometimes one bragur and sometimes another is possible with a stanza -- I mean this goes with that. Let us take verses like [he performs a III stanza]; then there is another variety [he performs two successive II stanzas]. All this had to be so variable that it was not really possible to learn from anyone. Now there is someone who did kvea in a different manner from this, either better or worse, and then it was a completely different impression. Therefore it was not easy to learn this; it had to be a little within people."
Karl not less than Thórur is aware of the diversity of kvæakapur and thinks he performs in a unique way, "not like any else." At the same time his stemmur of melodies are to all appearances derived directly from his father-in-law's melodies and seem occasionally to be close to the originals, as already pointed out. But in spite of not considering the melodies to stem from others, neither Karl, nor Thórur, nor any other performer in Breiafjörur claims that he composed them himself. In the opinion of an old informant the melodies are inherent in the verses since he "simply picked the melodies from the verses" but did not learn them from anyone.1 Even if the kvæaskapur Karl heard in his early years was not outstanding he had begun to practice when still a boy. Thórur and Gunnar tell the same story of practicing only without letting others hear them until they were grown up. They do not remember any special technical problems encountered, as for instance mastering the dillandi.. This ornament is apparently produced without effort by the performers of Breiafjörur with kvæa-voice, unlike the kvæamaur at the farm in Thorvaldsstair who had to shake his jaw as quoted earlier (p. 6/7).
One of the hands at Thorvaldsstair could imitate others but did not have his own stemma (kvæa-melody) like the rest of them. The same attitude appears here as in Breiaförur: each performed in his own manner unless he was imitating or parodying someone else consciously. Usually this was done for amusement rather than out of necessity, like the hand at Thorvaldsstair, or for demonstration like Thórur above. Thus the performers seem to be highly aware of the manner of other performers as well as their own without knowing the process of learning it. Singing is a different matter. Variety is present here too, but Thórur knows exactly from whom he learned each variant of the traditional hymn tunes he knows, and stylistic traits pointed out by him are recognized as regional characteristics. This attention to tunes existing in their own right is in striking contrast with the lack of it in his kvæaskapur, including the imitations of Snæbjörn, which are only successful in Thórur's opinion when he is in the right mood. Of the three imitations recorded only the first one is unusual melodically but nevertheless follows the outline similar to stanzas like No. 23b, prominent in his last performances. The remaining two imitations, recorded four years later, do not deviate markedly from his usual E (cf. No. 32a) and 'A' respectively. The voice is changed as in the first, except it is slightly higher in pitch, and traces of Thórur's ordinary kvæa-voice can be heard. Since the texts in all three are verses frequently performed by Snæbjörn, according to Thórur, the emphasis is apparently on voice and selection of texts. However, it is clear that unusual progressions in addition to complete change of voice contribute substantially to the illusion, created only in the first imitation, that the performer is different.
The first two points mentioned by Thórur in his remarks above -- variable voice quality and invariable bragur or stemma -- appear repeatedly in his comments on kvæaskapur. "It was especially the voice quality that made it [the kvæaskapur of Brynjólfur] more fascinating -- the variable voice quality affects the kveskapur." Since the quality of the kvæa-voice is essentially fixed even at the cost of intelligibility of the text, it is difficult to see Thórur's point unless he is using the term raddblær (voice quality) in a wider sense than is usual, so that variable pitch motion, ranging from inflections to sequences at the limits of the system, is included as well. Then a clear view of kvæaskapur as practiced by himself and others emerges. Variability belongs to the domain of voice, and the scope for "variable voice quality" is stemma, ideally fixed as "the bragur should really be the same."
Limits for variable voice are necessarily allied to the basic metres, which at the same time organize the verse metres proper in connection with alliterations and various rhymes within and between lines. Divisions of couplets into lines may be described in terms of two complete feet (4 syllables) shortened by 1 syllable or lengthened by 0-4 syllables with only the two longest (4+3 and 4+4) as the first line.

Fig. 13

A slightly different formulation results in only three types of lines (a, b, and c) which are either full or shortened by 1 syllable (-1 in parenthesis).


In the early period of rímur (ca. 14th through 16th century)2 all four varieties of c+a and c+b respectively in both couplets appear as variants in only two verse metres, úrkast and ferskeytt , later fixed as IId and Ia with regard to basic metre. The third combination c+c, on the other hand, is limited almost exclusively to 7+7 in stafhent (Ic) with 8+8 occurring in a few stanzas as the second couplet. It is natural that 8+7 (IIf not introduced until the 17th century) should not appear as a variant too, as it would conflict with the end rhymes (xxyy) characterizing this verse metre. If III is expected with IIf then the complete range of lines and couplets outlined above is present at the early stage of the metrical system without any clear distinction between I and II metres. Besides 8+5 (IIe) and ferskeytt limited to Ia the variants 8+6 and 7+5 were later used in independent metres, but no recorded performances of them are known. Úrkast was limited to IId, and IIc regarded as a separate metre; the remaining two variants, on he other hand, disappeared. In a couplet of úrkast (c+a) the whole range of lines is framed by a combination of the longest and the shortest. The range of couplets is similarly framed by combining c+a and c+c in braghent (IIa), a verse metre already existing in the earliest period. Afhent (IIb) has also c+a for a first couplet, but here the second line of the second couplet is omitted paralleling the second line of the first without a second member. Omitting the second member of a first line produces a new combination only with b; this possibility is therefore exhausted in the first couplet of Id. Thus the parallel structure of couplets and lines limited by 8 and 4 syllables is evident when the system is seen as a whole with Ia, the main metre throughout the history of rímur, exactly at the centre.
In Ia performances the correspondence between a first line of 4+3 syllables and a couplet of 4+3 feet is stressed by the usual 7 units of length in the first line, 6 in the second, joined to 8-9 in the third, and 6 in the last up to the seimur, totaling approximately 28 units. This pattern accords in fact with the basic scheme, apparently setting limits to verse metres and melody at the same time. The hierarchy of lengths involved may be indicated by an array of numbers:



If x stands for a foot of 2 syllables, and a refers to a line, and b to line-members, n to higher limit, and m to lower limit, the basic metres are formulated by fixing first lines and members at the higher limit and adding the optional -1 to each line. If, on the other hand, x stands for 2 units of length and a for primary and b secondary intervals, n narrow and m medium, the length at M, or the temporal ranges of the limits 1, 2, and 4 (1/2I, I, and 2I), are defined. It is clear too that temporal values of intervals are not only expressed at M but generally at all admitted weights if 0 < x < 2 units of lengths. Since frequency of occurrence is in direct relation to the temporal values and hierarchical order of the intervals (cf. Fig. 12), average interval use and the general properties of the metrical system appear as two aspects of the same structure.
The three positions for the first two alliterations with regard to the third, invariably at the beginning of the second line of a couplet, outline the range possible for the second line (cf. Fig. 14) besides fixing the first.



In this way the unique scheme of alliterations used in rímur metres delimits the basic metres in accordance with the scheme above (Fig. 15). But other systems controlled by the same scheme are conceivable; the basic metres should therefore be seen as a particular system of variants. Three of the earliest (Ia, Ic, and IIc-d) have parallels in foreign medieval poetry which may have served as starting points in conjunction with the scheme of alliterations already existing in the skaldic metre hrynhent .
Rhymes, on the other hand, fix the individual verse metres falling into the groups I and II in performance. The combination of a first line shortened by 1 syllable with a full second line of 8 syllables in III apparently does not lead to a I performance because it demands different phrasing. The general rule is therefore that a couplet containing a full line of 8 syllables, whether first or second line, is lengthened at the end. This rule is hardly possible without the regularity of verse metres attained in the 17th century. Therefore it seems likely that the same stemma was used for all metres in earlier periods when both types of first lines besides varying second lines could occur in the same ríma. Then a widespread practice today of using different tunes for different metres within the same group is perhaps the last stage of a process starting with the division into I and II performances.
In the "modern" practice each tune is simply considered a particular stemma and some at least are attributed to past performers. From this point of view Thórur is using his own stemma when imitating Snæbjörn, and Karl his father-in-law's when he believes he is using his own. Nevertheless the main point in imitating Snæbjörn is "that it is not possible to find the bragarháttur [stemma] he used or the kvæaskapur" in his biography or other written records. And Karl, on the other hand, seems not to be aware that his stemma ever changed. Stemma, as range for variable voice, is then reduced to the general metric and melodic constraints as manifested in the kvæaskapur of a particular performer. This manifestation could depend of various factors but probably in the first place on voice. A particular variable voice creates its own limits within the general limits and forms thus a specific stemma. Imitating someone's variable voice therefore gives an idea of his stemma even in a single stanza, as all stanzas limited by the same stemma represent it whether they are unique or close to a common tune. Karl may be close to his father-in-law's tune and perform nevertheless according to his own stemma formed through practice starting already in his youth. Thórur's I performances are unified by his I stemma in the same way, and the various melodic outlines or contours might partly be a result of not discarding previous possibilities when adopting new ones. But such alternatives, whether successive or simultaneous, besides other secondary factors could conflict with the stemma and eventually alter it. In extreme cases disparate elements would create the impression of changes of stemma or even lack of any definite stemma. Thórur remembers performers who changed the stemma frequently from one stanza to another. It is "an unfortunate blunder" in his opinion; he admits however that it could occasionally happen to himself when he is absent-minded and unable to concentrate upon the performance. Karl forgot the second couplet of the text of No. 30 for the same reason. He was absent-minded because he was making hay this day and wanted to return to work as soon as possible. The second half of No. 30a is reduced to humming the tune connected with his II and partly with his III stemma, as his variable voice is out of function.The pitches appear straight and stable on the melogram unlike those of the first half, and the intervals are almost the same as in his wife's F stanza No. 31a. At this moment Karl is probably trying to remember the text while not paying any attention to the performance, and when remembering it a little later, he gives a complete rendition in his usual manner (No. 30b). Gunnar's söngstemma, on the other hand, is associated with singing, i.e., absence of variable kvæa-voice. It is therefore not surprising that he knows exactly from whom and when he learned it unlike the kvæa-melodies or stemmur acquired in his youth. Gunnar sings the söngstemma with a timbre not far from his ordinary kvæa-voice, but volume is not constant, and near the end there is vibrato rather than tremolo on the long syllables not perceptibly inflected.
The two aspects of voice, timbre and pitch, were apparently Thórur's main concern the last time he was recorded. During a break in the performance of "Breifiringavísur," his favorite poem, he made this comment: "When strength fails everything is lost. Then it is not possible to get the tónar [pitches] as they should be; it is true because they should change a little according to the substance." And a little later he says that when he was younger, he could finish a text "without losing the right voice quality. It is as in singing, you see, that there is tenor and bass; it is of the same order but must be tuned differently...When the strength fails, it goes to the dogs."
This is the only time Thórur has used comparison with singing in order to clarify his comments of kvæaskapu . "Right voice quality" or timbre resulting from special tuning of the voice is no more attainable when the strength fails than the pitches "as they should be." That the pitches should be variable in accordance with the substance of the text could mean that frequently performed verses ultimately contribute to the specific constraints of stemma. It is probably not a coincidence that a performance of "Breifiringavísur" occasioned Thórur's critical comments. He has not only performed this text more often than any other recorded, but also cites a couplet or stanza from it occasionally in conversation. On such occasions the verses are spoken. Thórur, like most informants of the oldest generation in Breiafjöur and elsewhere, speaks the verses in an essentially impersonal manner characterized by relatively fixed, for the most part falling, intonations and regular articulation of the verse metre with short pauses between all lines. The general impression of this manner is monotonous, unlike modern recitation of verses, which is frequently individual and varied and tends to be interpretive as in other Western countries. It seems, therefore, unlikely that Thórur has interpretation in mind when connecting variability and substance in kvæaskapur, where the most diverse subject matters may be treated in a similar way and the same subject differently.
Before proceeding further a few points regarding the language of rímur must be mentioned. In Craigie's words a knowledge of the most common types of heiti and kenningar, "and the ability to recognize them under their infinite variety of form, is the first essential towards the understanding and true appreciation of rímur." Then he gives the following account:

The language of Icelandic rímur, like that of the old scaldic verse, is marked by the use of many words which are never employed in ordinary speed or prose writing, and by the frequent use of condensed similes which stand for the usual designation of a person or thing. In technical usage the former are called heiti (or more fully ókennd heiti ) and the latter kenningar ...Of heiti the most common are those which denote king , man , or woman ...Of kennings the commonest are those for man , woman , gold , poetry , battle , and sword .The explanation of these, which is often very recondite, must be sought in the Edda of Snorri Sturluson, a work to which the rímur-poets make very frequent reference. It is only through the infinite variety of expressions yielded by these kennings that the elaborate metres of rímur are rendered possible at all, and they are used for metrical purposes quite apart from any fitness in the terms themselves.4

Two such kennings, both denoting man, occur in the second and third stanzas of No. 14 for instance. The first, börinn kjóla, means literally "tree of ships," and the second, beygir geira, "bender of spears." Kennings are also made on other bases than skaldic poetry and the Edda but in a similar manner. Various paddings in the ordinary language were also used extensively and considerable deviations from the normal syntax tolerated. Thus a great range of possible expressions is available for any subject matter, and the same narratives were versified over and over again apparently for the sake of variety. Besides the Edda, a textbook for skalds written in the 13th century and only of use for rímur in connection with language, no theoretical writing existed until the end of the last century when the practice was already dying out. The stability of the tradition is in fact no less remarkable than its vast vocabulary, which was needed in order to produce the greatest possible variety within the framework of the metrical system.
Generally, d´yr verses (intricate and difficult, see p.9) were valued for their own sake. Thórur has said that the best stanzas are those "that can be read backwards as well as forwards" because in that way "they can really become two." And when he says on the same occasion that "it [kvea ] must be approximately be in right proportion to substance," the impression cannot be avoided that the sound of the words forms an integral part of the "substance" treated by variable voice in performance and that both are evaluated quantitatively, i.e., weighed. But estimation of substance could of course be influenced at the same time by subjective attitudes depending on personal life and environment.
In "Breifiringavísur" the poetess Ólina Andrésdóttir (1858-1935) describes the daily life in Breiafjörur, both work and entertainment, as it was still in Thórur's youth. It is a simple description without any novel ideas or observations but cast in a not too easy variant of Ic with well-fitted words. The 12th stanza most often quoted by Thórur:

Vetrar löngu vökurnar
voru öngum thungbærar,
vi ljóasöng og söngurnar
söfnuust föngin unaar.

is loosely translated

The long winter evenings
were not hard to bear;
listening to songs and the sagas
brought joy.

but a more exact rendition of the second couplet would be

by songs and the sagas
stacks of pleasure were gathered.

The kenning in the last line "stacks of pleasure," alluding to the stacks of hay gathered in summer during the sunshine, has especially been pointed out by Thórur, who has frequently said that kvæaskapur brought light and warmth into the ill-lit living room on cold winter evenings.
The particular significance of this text to Thórur clearly affects its performance characterized by exceptional intensity and personal involvement. Such performances tax his strength, as he is panting after finishing each stanza; hence the unusually long pauses in No. 11. It is also noteworthy that A contours are used exclusively in the third (No. 11) and fifth of the five performances investigated but not in other performances of comparable length. Generally there seems to be no relation between texts and use of the various contours in Thórur's I stemma. However, it is clear that Cw appears especially in performances marred by misspellings because Thórur is not well acquainted with the text he is reading or not concentrating well for other reasons. But since he is trying his best in the two A contour performances, sticking to the main contour is possibly a result of great concentration even if failing strength makes the latter unsatisfactory in his opinion. No exceptional stanzas appear in this performance, and weight between stanzas as a whole is considerably lower than in the third, where it is unusually high (between 15 and 30). Weight and unexpected rather than specific progressions seem in fact to be Thórur's main concerns regarding substance. Timbre, on the other hand, is fundamentally constant both with respect to "tuning" and volume and seems to betray emotions only exceptionally, as in the third performance, where the voice is narrower still than usual and sound sometimes as if Thórur were clenching his teeth. Such a performance could be called weighty and the easy ones light. Thórdur's Cw performances are then his lightest, reverted to when he has difficulties in concentrating, and probably at the outskirts of his I stemma if not outside it because of the resistance of Cw to change. The third performance of "Breifiringavísur," at the other extreme, and the last attempt at such a performance are on the contrary limited to the main contour not only of his stemma but of the whole Breiafjörur region. Elsewhere in the country A contours are rare even if a number of diatonic melodies are related to them in various ways, and since they are not used by the younger generation in Breiafjörur either, they apparently serve variable voice and disappear with it.
The position of Ia in the metric system is comparable to some extent to the A contours. It exists in more variants than any other basic metre, ranging from simple, frequently referred to as light, to the most complex or d´yr. In extreme cases the subject matter is negligible, and the substance of the verses turns into almost pure sound.

Máttur réttur snjalla snill
snara færa kynni.
Háttur sléttir valla vill
vara tæra minni.

This stanza is the last of the 54 in the metre sléttubönd making up the 17th ríma of Rímur af Flóres of Leó by Rev. Hallgrímur Pétursson (d. 1674). It can be changed in 48 ways without changing the meaning or the metre, an especially d´yr variant of sléttubönd, and since sléttubönd can be read backwards as well as forwards the variants are altogether 96. Performing the stanza in all ways would expand it to the length of a ríma without adding anything to its content. It is not known if this has ever been done, but it is possible, as performances prolonged simply by repeating the same stanza or part of it over and over again were not unusual. Björgvin Alexandersson, Gunnar's brother, remembers a foreman prolonging only one stanza while steering his boat homewards by repeating the third line until he reached the shore; then he finished with the last line. Karl, on the other hand, repeated a complete stanza a number of times in order to keep warm while watching sheep in winter. Total length is of major importance too in ordinary performances without repetitions. Gunnar tends to slight performances of one or two stanzas; several stanzas at least are needed for an acceptable performance. If a weighty performance is characterized by kvæa-voice variable to a considerable degree (weight between stanzas above 15) and well sustained through a sufficient number of stanzas (the first four, No. 14) from Rímur af Atla Ótryggssyni (Atla-rímur) are no less weighty than Thórur's "Breifiringavísur" (No. 11). But the means by which similar total weight is achieved are obviously different.
As described earlier (p. 41) the relation between the third lines of the third and fourth stanza of No. 14 consists of five operations. They are all of similar weight and result in a total weight of about 9.5 (i.e., 2+2+2+1.5+2). The total weight of the three operations between the third lines of the first and the second stanza of No. 11 is only slightly higher, 11 (0.5+2+8.5), but the last operation at M

stands out from the first two, (+1-1) and (-2+2). If the operation of one or two pairs are considered simple and operations of three pairs and more complex, Gunnar's operations are all complex except the fourth, whereas Thórur's first two ((+1-1) and (-2+2)) are both simple and only the last complex. Additionally, Thórur's sudden and relatively great expansion of the total range should be noted. Gunnar's line, on the other hand, is compressed as a whole in a similar way to when (-1+1) is applied to a whole line, for instance in Thórur's performances, resulting in similar weight. Gunnar's manner is therefore like the usual syllabic practice with regard to general distribution of weight and expansion or contraction of range even if the means are many complex operations of small weight each. The relation between the second lines of Kristján's No. 7 is in some respects intermediate. Three simple operations, (-2+2), (-4+4), and (+2-2) of total weight 5 (2+2+1) are followed by a complex operation of six pairs beginning on the long syllable and ending on the third syllable of the last line:

Since its weight is at least 6, weight and complexity increase suddenly between the lines at the end, as in Thórur's case. But the weight is more evenly distributed on the whole, and the last operation resembles Gunnar's last especially if it is kept in mind that Kristján's eighth-notes are of the same absolute length (about 0.15 seconds) as Gunnar's sixteenth-notes.
Complexity of operations of a given length whether within or between stanzas depends evidently on the rate at which changes of pitch occur. The modalities narrow, medium, and wide, and intermediate stages, may be regarded as resulting from different rates of change with the hierarchy of intervals discussed earlier when the average number of intervals (including zeros) remains the same for all stages. A high rate increasing average number of intervals leads to ornamentation on the other hand, without necessarily changing the modality. In Breiafjörur wide characteristics are only found in the north (Barastrandars´ysla), the only part where heavy ornamentation is completely absent. But wide modality appears there only sporadically and in Thórur's performances not clearly except in the second and third line of No. 23c, his first imitation of Snæbjörn. The pitch changes (or non-zero intervals) in the two lines are 15, but only 10 in the corresponding lines of No. 23a and b respectively, where the total number of intervals is nevertheless almost the same as in 23c. Generally, widening does not lead to greater complexity of operations within stanzas, as the limits would soon be exceeded. The complexity in No. 23c and of b, where "b-flat" is r until the beginning of the second couplet, is similar and so is the total weight, near 4M, and its distribution.
The value of maximal total weight within and between Ia-c stanzas corresponds approximately to the average number of intervals, 32, in syllabic performances. But the number depends on the interplay between rate of change and articulation of metre. It is possible to vary this interplay endlessly without changing the average interval use or modality, if it takes place within fixed limits. The basis is apparently the total length of a stanza with seimur, consisting of 32 units of length on the average, varying ±8 if all metres are included except IIg and III. In most cases the total length is even more stable in the same performance than stanza length without seimur, both with regard to relative length varying at most ±2 units and absolute length ideally delimited by a single breath.
One should note now that the word lota, sometimes used for seimur, means in the first place in current Icelandic usage the time-span of a single breath of something done without a break. The term lota when referring to kvæaskapur could therefore denote at the same time a fixed time-span and pitch level. Then a lota of 32 units, the length of the interval 0.25 at M, defines a zero pitch level variable by 0.25 at the same time as the 32 units constitute a fixed reference or zero level in terms of length. From this follows that all intervals are variable ±0.25 and that ambiguity sets in only with 0.5, as assumed earlier.
A lota of 32 units of length and the 32 metric syllables forming the basis for the verse metres are two aspects of the same reference level with regard to the basic scheme (Fig. 15) and lead to similar limits for total weight, as the concept of weight may be applied to other dimensions of sound than pitch.5 If each rhyme and alliteration weights the syllable carrying it by 1, the lowest weight for four-line metres is 10 (6 alliterations + 4 rhymes) and the highest rarely more than 32, reached in Ia when all syllables rhyme including the 6 carrying alliterations (cf. Hallgrímur Pétursson's stanza above and the last variant of Gunnar's couplet). However, in a few instances like the second variant (Ic) of Gunnar's couplet, the weight is 34, and the first variant (IIf) exemplifies how it can increase to 36 in exceptional cases, and the highest value theoretically is obviously 38 for IIg. The lower limit, on the other hand, is brought to 7 (i.e., 5+2) in Id and to 9 (5+4) in IIb, generally considered the lightest of metres. But most of the four-line metres in the texts performed fall into the range 12-16 at least. Weight indicates here a change from the basic metres as it indicates for pitch a change from reference levels within stanzas and from other pitches between stanzas. Gunnar therefore slights the lightest metres possibly on the same ground as he slights short performances. In some respects, however, d´yr verses and variable pitch lead to opposite results. An increased number of rhymes diminishes the variety of sound within stanzas and restricts the choice of words. Metre is therefore in certain measure comparable to kvæa-voice, which diminishes oppositions between speech sounds. Both are in fact essentially constant in relation to variable pitch on one hand and wording on the other, as metre is variable in the first place with regard to the metric system as a whole.
Performance, unlike verification using the discrete units of language as material, is assumed to be based on a pitch-time continuum delimited by lota and subdivided according to the basic scheme and other limits of stemma by the performer's rate of change, never in exactly the same way. Variable voice in this context can be defined in terms of the continuum in relation to the variable rate of change joined, at the level of inflections, to the continuum and to constant timbre, or constant "tuning," and supply of breath controlling lota at the same time. Accordingly, the variable division of the continuum is conditioned at every moment by constant timbre and volume as well as stemma, and the resulting measured sequences (operations) are therefore "variable voice quality" no less than a particular instance of stemma.
It is now easily seen that secondary references or limits of stemma approaching a fixed tune correspond to a division variable within the narrowest limits if the pitches are at least inflected. If they are stable the division is constant or nearly constant. But in that case constant timbre is usually replaced by more variable timbre approaching the singing voice and attention turns from division to the tune fixing it. In kvæaskapur, on the other hand, not tunes but momentary changes attract the interest both of the performer controlling them and the audience. According to Thórur, it was really "only these changes in sound [of the voice] that were admired" in the performer's art which "is invention but not education." Then he adds, "But invention is education too, isn't it?"6

Even if only a few aspects of the performer's art have been investigated, it is clear that it is highly intricate and elaborate in a way comparable to the poet's art. It is clear too that performance does not serve the text directly any more than the form of the text serves the subject matter. However, the oppositions and interrelationships between the various aspects of performance and verification bring the listener ultimately to a heighten awareness; or simply in Pétur's words, "The narrative becomes more impressive if one performs beautifully." Thórur has stressed the same point and added that only a good kvæamaur had "such an influence that one could repeat every single word of the verses," otherwise the audience was less attentive. Thus impressive delivery that holds the interest, but not clear articulation, is without doubt what Pétur had in mind when defining the performer's art essentially as good delivery of the text. The cumulative effect rising from all factors involved may be indicated as in Fig. 17, where 1 stands for the relations between subject matter and verification or substance of verses, 2 for the relations between variable voice and verification, and 3 for the total effect. But the numbers can also stand for bragur in three senses: 1 as verse metre, 2 as stemma, and 3 as the fundamental organization underlying both and approximated through Fig. 15 and the principle of a fixed reference (lota in performance).



From this point of view performance and verification (consisting of wording and rhyming) are complementary, and the poet depends no less on the performer than the performer on the poet. Accordingly, a poet composing a text fit for performance has not only substance 1 in mind, but also 3, the ultimate effect of his verses, since knowledge of bragur 3 and the general performance practice enable him to estimate the right balance between substance 1 and 2, even if the latter can be realized in countless ways. But since performance is expected to be individual too, individuality forms part of this balance. Snænbjörn and others who performed "as if they felt their importance" apparently went too far in Thórur's opinion by exaggerating their idiosyncrasies. Verification, on the other hand, is not individualistic, at least not in any obvious way, and the poet was expected in the first place to be a superior craftsman according to common standards. Individual or specific limits belong then only to bragur 2. The general limits of bragur 1 and 2 intimately connected and stabilized by 3 are nevertheless subject to influences changing them in the course of time. but since the specific limits have been formed during the course of a lifetime, bragur 2 belongs to the personal history of the performer as well as to the history of the system as a whole.




1 Gisli Marteinsson, Baraströnd (b. 1887)

2 Historical information on early rímur from Thórólfsson (1934) and (1950), and Craigie (1952).

3 In this connection it is of greater interest which metres were selected than where they came from. The later question is controversial. Vésteinn Ólason (1976) has pointed out that it is reasonable to assume that Icelanders had some knowledge of the narrative poetry practiced in Germany and England in the 13th and 14th centuries which rímur resemble in certain respects. Thórólfsson (1934) and Craigie (1952) maintained, on the other hand, that the rímur metres in question had their origin in the medieval Scandinavian ballads.

4 Craigie (1908), 4 and 104-5.

5 Weight for any dimension of sound means always a quantitative measurement of a distance x of a given length 1 expressed xl as in (7) for pitch. It is then clear that the weights for different dimensions are not commensurable unless the references are, as in this case.
Counting only correspondences between speech sounds, each alliteration and each rhyme is simply one correspondence signaling out the metrical syllable carrying it. Since only this property of sound is taken into account in this connection, the difference or the distance of all syllables sharing it from the syllables of the basic metres (sharing only uniform length) is the same, except when a syllable is signaled out by the two schemes (alliterations and rhymes); then the distance is doubled. The value of this distance is conveniently set at 1 in accordance with the mean value of a small second arbitrarily chosen as a measuring unit for pitch distance or interval value.

6 In the letter quoted by Bjarni Thorseeinsson (see p.1) Benedikt Jónsson makes a comment relevant here. It is translated from the original (dated 1899) in Stofnun Árna Magnússonar á Íslandi, Reykjavík: "The true rímur-melodies lack much more in form and definition [than the fixed kvæa-melodies] so it is hardly possible to catch them; most of them are unmelodic and little else than a ringing or singing declamation with long 'lotur ' ...[The performers resort] to a few tones only (most often a major or minor third) but compensate for the poverty of the tones by dynamic artifices, change in sound [hljómbreytingar, the same word as Thórur uses], accents, graces, trills, and all the varieties within the capacity of the vocal organs...In this way rímur when performed here [i.e., in Thingeyjars´ysla in the north east], and I am convinced that some of the kvæamenn exercised a considerable and far from despicable art by this."