What Olof had in mind Herschend, Frands Fornvännen 87, 19-31 http://kulturarvsdata.se/raa/fornvannen/html/1992_019 Ingår i: samla.raa.se
What Olof had in mind
By Frands Herschend
Herschend, F. 1992. What Olof had in mind. Fornvännen 87, Stockholm. An investigation of the characteristics of the weight of the Sigtuna coinage as published by Malmer 1989. The firsl coins, although reflecting several norms, are considered to have been deniers defined as centred around the weight of 1/96 mark of about 208.5 g equal to 16 smaller units. The mark is the same as the one governing the Gotlandic bracelets of Stenberger's type 2, Herschend 1987. With the New Series the symmelrical ideal was abandoned and the average coin weighl dropped since the coins weighing between 12 and 14 small units came to dominate the material.
Frands Herschend, Department of Archaeology, Uppsala University, Gustavianum, S753 10 Uppsala, Sweden.
The reconstruction of the Viking Age metal weight system on Gotland has made substantial progress during the last 20 years, as proven by studies such as Lundström 1973; Kyhlberg 1973, 1980 and 1982; Saers 1982; Sperber 1986, 1989 a and 19896, or Steuer 1987. It is therefore only natural when a contemporary coin material from the Mälar Valley, although mainly found on Gotland, is published to look for affmities between the two regions. Such and wider connections have of course already been discussed, e.g. by Kyhlberg (1980) or Sperber (1989 a). This is a promising lane of research whereby this artid e seeks a connection between Gotland and Sigtuna, based on the splendid publication of the Sigtuna coinage c. 995—1005 (Malmer 1989). The discussion of weight in Malmer 1989 is preliminary within a chronological perspective, pp. 33 f. Its theoretical base is an assumption that the coinage mirrors definite weight systems, while the methodical approach derives from a comparison of minimum, maximum and mean weights in different subsets of the coinage. As a framework for the initial coinage the reader is offered a comment upon our present day knowledge:
English coin weights can hardly have set the standard for the Crux period of lhe Sigtuna coinage. Was there another model for its high average weight? Or did Sigtuna set its own weight standard independently? The weight systems in the Baltic region during the Viking period cannot be easily discovered despite lhe many balances and weights thal have been preserved (Steuer 1984, 283-6). At the present time it is hardly possible to determine the early weight standard of the Sigtuna coinage in terms of, for instance, the weight of a mark or its subdivisions. The most . .. (Malmer 1989 p. 32.) The chapter on weight consists of a preliminary presentation of the material and a discussion which implies that some of the opinions held in Malmer 1965 and Petersson 1969 may still be valid, if the analysis, aided by the comparison of minimum, maximum and mean weights, refers to a well structured numismatic material, in which the often small subsets are defined by several intricate numismatic variables. In the discussiem of the weight of the Crux imitations pp. 30 f. only some 12% of the coins are mentioned. They fall into two subsets, 28 and 13 coins, and it is observed that their weighl relation is close to three to four. The smallest set is grossly influenced by two extremely heavy and two rather light coins. If we removed just one of the heavy coins the mean weight of the sample would
Fornvännen 87 (1992)
20
F. Herschend silver, since they can obtain prestige goods without taking an active part in risky trade expeditions or piracy. They must of course, a.s their part of the cleal, pay for their goods, which is a two-fold problem. First they pay for the entrance to the märket and then they buy at relatively stable prices. The entrance fee would be a real weight payment which may well have been linked tei the duty of using only the King's money on the märket. Tei reconstruct the principle we may picture the following scene: The peasant puts one mark eif silver onto one of the pans of the official scales, in order to halance the official mark weight in the either pan. Then the weight is removed and the number of (mixed) coins officially equalling a mark are piled in the pan, to balance lhe silver. If the coins match the silver, then a certain lunnber of coins, perhaps a eighth, are removed together with the silver, by the officials. The rest of the coins are given to the peasant, who will now know what he has really paid for his money and eventually for his goods. Whatever fees may have been imposed on those who wanted to enter the märket, in order to buy, or to sell, or to establish a production, they must have had to pay a real value charge. The point is of course that for people in the region it is an absolute rather than a relative economic advantage to be on the märket. Therefore the King may also favour you in different ways, payingyou with his money or giving you a site in his märket town. Between märkets the value of the transaclious is always a matter of the real values, (the weight eif the silver) and for this reason you cannot, when you create a controllcd märket in a region with an established tradition of a real silver weight economy, very well defme the price of your coins on nominal grounds only. O n the contrary the silver content must be high, and the mean weight a natural fraction of a mark. When the märket expands into the region around the town and when the hoarding of silver has come to an end, then the need for simple metrology dedines. Instead it becomes essential that the coins are so over-valued and standardised as to weight and
d r ö p from 2.46 to 2.35 g, and if on the other hand one of the light ones were substracted the mean would rise to 2.53 g. O n e third of all ( a u x imitations are made with a single die, but the mean weight eif these coins seems to be irrelevant to the discussion, although it is an average weight not significantly changed by the loss or addition of an extreme weight. Bearing the inner structure of the chapter in mind the discussion seems very appropriate, but readers having an eye to metrology must resort to the catalogues and create a database of their own, bearing in mind the numismatic structure of tbc coinage so splendidly scrutinized in the book. The analysis must be based upon large subsets, and should not rely solely on rough measures, such as the mean in grammes down to two decimal places. Using this approach the analysis will hopefully disdose the general metrology while recognizing tbc complexity of the Viking Age economy and its currency. A theoretical point of departure When, as in this Mälar Valley case, local market-dependent money is introduced, then the weight of the coins is likely to be related to the metrology already existing in the area, since this system has hitherto been used to defme prices as ecpial to weight, in a fair but probably slow silver weight-dependent trade. The great achievement does not consist in mastering the basic techniques of coin production: on the contrary, it is the ability to create a märket on which the neiminal value of the money is stable rather than floating, and thus not so strongly bound tei the real silver value of the coin. Sigtuna is in my opinion such a märket. During the Viking Age a silver surplus accumulated in the Mälar Valley mostly due to peaceful or armed initiatives taken by the most influential or well-to-do peasants of the region. This surplus was one of the reasons why the King created a märket. His contribution was a safe märket place visited by foreign tradesmen, bringing attractive goods into the town. Such a märket is particularly advantageous to peasants with modest fortunes iu
Fornvännen 87 (1992)
What Olof had in mind silver cemtent, that only the few who trade between märkets are interested in their real value. To begin with, when Sigtuna is the only märket for the coinage, then the weight and the quality of the silver is more important than the quality of the mintage and the design of the coins. The representative sample
21
The first sample The round coins in Malmer's first catalogue, Classic series, is the eibvious first choice if you want tei study the initial coinage. The coins included in the sample are those with no comment in column 10, n o r comments in column 5 implying a loss or an addition of weight, and moreover no ceiin with a crack eir a hole (column 13). Coins of the last-mentioned quality are included in the sample used by Malmer 1989 when describing the weight of the subsets mentioned above, but with the approach of this case-study, the fact that cracks and holes are secondary traits, on average lowering the coin weight, is enough to e x d u d e them from the sample. The effect is best seen in the 271 coins of chain 1 used in Malmer 1989. They have a mean weight of about 2.14 g, the 239 coins with no crack or hole weigh 2 . 1 7 g and the 32 coins with cracks and holes only 2.09 g. For chronological reasons one should also e x d u d e the coins of chain 10 from the initial coinage (Malmer 1989 p. 23 ff). There are 297 coins in this sample, 239 belonging lo chain 1, 39 single coins and 19 in the small chains 2—6 and 17, i.e. 58 coins outside chain 1. The coins of chain 1 no doubt belong to the intial etiinage, they may have been produced during a two or three year period in the late 990's. The coins of the small chains and tbc single ones may or may not be a part of that coinage, since their dies are not, at least not yet, linked to chain 1. Perhaps they are a låter or a late part of the coinage (e.g. the Long Cross imitations in chains 5 and 17) if not unofficial coins (e.g. the single ones). Obviously even a small chain may be part of an unofficial coinage, although several of these will probably link up eventually with a larger chain of official coins.
The sample handles some problems of representation (e.g. the problem of unofficial money) without leising too many coins; but at least two or three problems should be discussed since they have no eibvious solution. They are reflected in the questions; Was the weight of the coins influenced differently by the soils in which they spent between eight and nine hundred years? Have the economic factors, such as coin sort ing by weight, left us with a biased coin weight distribution? There are no definite answers to the questions. Concerning the first, Metcalf (1987) gives an example of differences in weight obviously due to soil conditions. It has, however, also been shown that coin sorting by weight may result in an intricatc weight pattern correlated to the size of the hoard, even theiugh the find circumstances and the quality of the coins are homogeneous, within a defined economic region, as is the case with the Oriental coins feiund on Gotland (Herschend 1989). O n e must, moreover, be aware that the weight of the Sigtuna coins may even vary with distance and time from their main märket and period of production. Coins are more likely to be hoarded outside their main circulation area than within it, especially if the region outside the area has no märket with controlled prices. O n e should in either words not use coins from Denmark, Gotland or the Mälar Valley only, even though we cannot possibly know what linds will best mirror the original weight distribution. For this reason, in order to level out oddities, one should bring all types of hoards and stray linds from different regions into the sample, although single coins found in town layers are probably in a poor condition compared with a hoard discovered in the welldrained calcareous soil of a Gotlandic farm. In this case it was wholly impossible to conduet a study of weighl in the light of the state of preservation, and thus only the demand that the coins be very well preserved can be said in a general way to e x d u d e unsuitable coins from lhe sample. Certainly, all told, the demand for differentiated finds is in conflict with the demand for
Fornvännen 87 (1992)
22
F. Herschend
. 2 2
0
råa ra va vä m
Fig. 1. The weight of the 297 coins from the Classic Series, Malmer 1989, chosen as a first sample for weight analysis of the intial Sigtuna coinage. - De 297 myntvikterna ur Classic Series, Malmer 1989, som bildar del första urvalet för alt studera Sigtunamyntningens första fas.
Fig. 3. The relative weight of the 239 coins in chain 1. The position of the median weight below the mean of 100 is indicated by the point where the cumulative curve meets the 50% line. - Den relativa viklen hos de 239 mynten i kedja 1. Att medianviklen ligger under genomsniltsviklen indikeras av den punkl där den kumulative kurvan skär 50-procentslinjen.
only well preserved coins. Therefore what matters is the weight pattern dissolved by the analysis of large samples. If the pattern is not relatively simple, no interpretation can be made of it. If on the other hand the pattern is simple then we are entitled to say that neither has the state of preservation distorted the material nor have the oddities caused by economic factors blurred the general picture. Small subsets, however precisdy defined, may be, and probably are, significantly influenced by
an overwhelming number of unkneiwn factors, making even simple patterns difficult to interpret. The principle of the reasonable weight distribution of the large sample is well illustrated in Melcalf 1987. In a wider perspective the problems touched upon here refer to the general source critical problem of reconstructing intent. The first interpretations The diagram of the basic sample, Fig. 1, shows a slightly skew weight distributiem. If the material is divided into the coins of chain 1 and those outside, i.e. the unquestionably initial coinage and the probably initial coinage, then the picture changes. First of all the mean weight is split. Among the 297 coins of the first sample it was about 2.11 g, but the 239 coins of chain 1 have an average weight of some 2.16 g, while the mean of the 58 other coins is only 1.93 g or so. Grouped in relation to their mean weight and compared with each other the initial coinage of chain 1 shows itself to be much more symmetric than the single coins and those of the small chains, Fig. 2. The distribution of the unquestionably initial coins, Fig. 3, is still slightly askew, but it is not wrong to assume that a fair part of it was once intended to be symmetric around a weight close to, but due to the skew-
%Mchsut»et= 100
J
56
, Qa
156 166 125 136 145
Li
85 75
i L i J U _ „
96 106 115 wwghl mean • 100
Fig. 2. A comparison based on lhe percentual distribution of lhe relative weight of the coins in the first sample, see Fig. 1, divided into two subsets: the coins in chain 1, hatched bars, and those outside chain 1, blank bars. - En jämförelse baserad pä den procentuella fördelningen av det första urvalets relative myntviklcr, se Fig. 1. Mynten faller i ivå delmängder: mynten i kedja 1, skrafferade staplar, och mynten utanför kedja 1, ofyllda staplar. Fomvännen 87 (1992)
What Olof had in mind
23
08/1,08 10/1.36 12/183 14/1.90 16/2.17 18/2.44 202.71 22/299 24/3J26 uniB/gr
27
Fig. 4. The absoule frequency of the 239 coins in chain 1, distributed among weight classes equal lo 1/1 536,h of a 208.5 g. mark, i.e. 0.136 g. - Den absoluta vikten hos de 239 mynten i kedja 1 fördelad på viktklasser som motsvarar 1/1 536 av 208,5 g d.v.s. 0,136g.
ness a trifle heavier than the mean weighl of 2.1659 g. Guided by these facts the next step must be to translate the abseilute mean weight into a weight unit. This is where the silver weight metrology frenn Gotland proves useful. Two mark weights have been defined with reasonable statistical significance, 223.8 g and 208.7 g. Both weights are related to Arabic and further off Roman metrology (Herschend 1987 or Sperber 1989 a, b). The bigger mark contains 103.32 mean weights and the smaller one 96.35. Knowing that the mean weight is slightly below the ideal, we come d o s e to describing the marks as containing either 103 or 96 mean weight units. The latter fractions are the more natural ones when il comes to coin weights since arithmetically 1/96 of a superior weight is the classic definition of a denier. It is thus more than plausible to understand the central weight sought for as a natural fraction, 1/96, of a mark of about 208.7 g. The probable central weighl There are three steps to follow in order to lind the optimum central weight. You start by defining the weight. Then you divide the central weight into fractions which in their turn defme the weight classes of the diagram, at last you check the distribution for symmetry in order to convince yourself, though in this case there is no hope of making the whole distribution symmetrical.
The weight area of the distribution in Fig. 3 is large, some 2.6 g. This is a good reason for understanding the distribution as a result of the coins belonging to different classes and it means that not all coins were considered to weigh the same. None the less the second point mentioned above poses some problems. On what empirical grounds and how should one decide upon a certain dass breadth? This is obviously to a great extent a matter of finding a new supporting context. Luckily the small fragments of Oriental coins in the weight purses of the late lOth century boatgraves Valsgärde 12 and 15 (Lindqvist 1956 p. 18) offer a straightforward contextual solution. They clearly show that the meist fréquent adjustments of coins weighed between 0.08 and 0.16 g (Herschend 1987, p. 190, Fig. 12). If the Viking Age peasant, when testing the weight eif a coin detected a difference of more than about 0.12 g, then he would probably not consider the coins to be of the same weight. A dass breadth of about 0.12 g is thus not unreasonable. This means that either 1/16 or 1/18 of the mean weight is to be preferred as the width of the weight classes. A test shows that a dass breadth 1/16 gives a next to one coin symmetrical distribution among the seven central classes if the central weights is equal to 1/96 of a mark weighing between 208.4 and 208.5 g, Fig. 4. The fraction 1/18 does always lead to an inferior result. This is only natural, given the fact that the coins are designed to be easily divided into four parts. Due to the fact that the distribution must still be slightly affected by a small loss of weight ameing the coins, since they have at least to some extent been used, I choose the highest value (208.5/96) to designate the central weight, i.e. 2.171875 g. The difference between this and the value which can be obtained from Gotlandic bracelets (Herschend 1987 p. 188f.) is about 0.002 g, and in my opinion negligible. The initial Sigtuna coinage seems to be designed to be a symmetrical distribution in which the central dass of coins may well have been understood as equal tei 1/96 or 16 smaller units of a local mark weighing about 208.5 g. If we check Malmer 1989 for the weight of
Fornvännen 87 (1992)
24
F. Herschend coins in the New Series fall in two big chains, 10 and 11 (192 coins), a dozen small ones (42 coins) and a few (4) single coins. The latter are most often square ceiins and weakly represented among the reiund ones. The connection between the two series is made u p at least by the coins of the chains 5 and 10. It may be argued that the division intei series gives a false impression of an abrupt change which was in reality gradual, but a comparison between the early chain 1 and lhe låter chain 10, or chain 11, or the late Leing Cross imitations with blundered obverse legends starting with an O, shows the same type of contrast as the one between the classic and new series, Figs. 5 and 6. Although the old central weight of 16 units is still visible the main classes have shifted from 15-17 to 12-14 units and lhe mean weight has d r o p p e d from about 2.17 to about 1.72 g. In chain 1 4 4 . 4 % of the ceiins could be found in the three main classes, but in the samples of Fig. 6a—c there are 59.7, 57.5 and 6 3 . 3 % respectivdy in these classes. The standard coin is thus becoming lighter but also more standardized in an asymmetric weight distribution with only about 15 % of the coins above the three most fréquent classes instead of some 2 5 % in chain 1. There is nothing to indicate that this change was gradual rather than abrupt, and as a paralld tei the material in chain 1 in ihe Classic Series it is reasonable to say that the material in the two big chains of the New Series sheiuld be chosen to show the contrast between the coinages which circulated at the end of the first millennium and the beginning of the second, Fig. 7. The character of the symmetry The weight classes eif 20 and 21 units in Fig. 4 are more likely tei have belonged to an original ideal of symmetry than the classes 11 and 12. The frequendes of the last two, teigether with thal of the classes above 21 units seem rather to mirror a wish to create a slightly asymmetric distribution. In these classes one may say that ceiins have been added tei an originally symmetric ideal. Before the analysis cemtinues it must be decided whether or not the distribution of
ni il II
E"ig. 5. Tlie weighl of the 238 coins from the New Series, Malmer 1989, chosen as a second sample for weighl analysis of lhe initial Sigtuna coinage. - De 238 myntvikterna ur New Series (Malmer 1989), som bildar del andra urvalet för att studera Sigtunamyntningens första fas.
the coins with a specific obverse die, then the symmetric ideal does not link in with the frequent dies. The weight distributions in connection with the dies do, heiwever, differ considerably, while at the same time they match each other. Even if an obverse die seems to be connected with light coins, e.g. number 13, the picture can be complicated. This obverse is nearly always found with the reverse 5 1 , which has in its turn been used together with nei less than six different obverses and the heavy "sceut" coins, i.e. the small subset of 13 coins discussed by Malmer (1989 p. 31 f ) . As pointed out by Malmer 1989 the quality of the coinage improves from chain 1 tei chain 11. That is to say inter alia that there is a more rational balance between the decreasing number of obverse dies and the growing number of reverse dies, the dies are more evenly used, the tendency to use a reverse die solely with one obverse is a little meire noticeable and the links between the obverse dies consequently fewer. As compared with the cemtemporary Danish chain published by Blackburn 1985, the Sigtuna coinage is, however, still not very developed. The New Series The difference between the weighl distribution of the Classic Series and tbat of the New Series consists in a change of proportions and a general loss of weight, Fig. 5. The 238 round
Fornvännen 87 (1992)
What Olof had in mind
25
,
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Units
27
Fig. 7. A comparison between the 239 coins from chain 1 (hatched bars) and the 192 coins from chains 10 and 11 (blank bars). - En jämförelse mellan de 239 mynten frän kedja 1 (skafferade) och the 192 mynten från kedjorna 10 och 11 (ofyllda slaplar).
Fig. 6.a-c. A comparison between I lie 297 coins from the Classic Series (hatched bars) and lliree different subsets from lhe New Series (blank bars) Suliset a: the 72 coins of chain 10. Subset b: the 120 ceiins of chain 11. Subset c: lhe 98 coins of the Osivle-. Malmer 1989 p. 171. - En jämförelse mellan de 297 mynten från Classic Series (skrafferade staplar) och tre skilda delmängder från New Series (ofyllda staplar). Delmängd a: de 72 mynten i kedja 10. Delmängd b: de- 120 mynten i kedja II. Delmängd c: de 98 mynten i O-Stil (Malmer 1989 p. 17f).
chain 1 is normal. There is of course a variation among the coin weights which should be attributed to chance, bul can tbat variation be responsible for lhe distribution between the seven central classes?
Using the mean weight and the standard deviation of the 195 coins of chain 1 which fall in these seven classes it can be ceimputed how the coins would have been distributed bad they been part of a normal distribution. The difference between the expected normal and the observed norm can be studied in Fig. 8. The observed distribution is not normal if you test it with a /"-test, nor is it significantly different from a normal distribution in the seven classes. Although the material is too small to show a significant deviation from the normal distribution there are d u e s to understanding the principles behind the symmetry. First ofall it seems as if the proportions between the classes 15—21 aineing the chain 1 coins are similar lo ihe proportions of the classes 8—14 among the round coins of chains 10 and 11. O n e might say tbat they constitute the upper and the lower seven classes of a symmetrical distribution comprising 11 classes. This is especially clear if you compare tlie percentage distribution of the twei subsets as in Fig. 9. The diagram shows well above half of the lower and upper pari of two distributions of the same character. The upper classes are taken from a symmetrical distribution, namely the initial coinage. The lower half, which is made up of the round coins in chains 10 and 11, has never had all its symmetrical counterparts, onl) some few coins in the classes 15-19 and 23
lArnviiiiiirn 87 (1992)
26
F. Herschend
Abs nos
48
36
24
12 obs exp
12
14
16
Units
18
20
Fig. 8. A comparison between a normal distribution, defined by the mean weighl and lhe standard deviation of the observed distribution, i.e. the 195 coins of chain 1 in the seven central classes of Fig. 4, and the observed distribution of the coins them-
sdves. - En jämförelse mellan en normalfördelning definierad av genomsnittsvikten och standardavvikelsen hos de 195 mynten i kedja 1 i de sju centrala klasserna i Fig. 4 och myntens observerade fördelning.
comparable to the coins of chain 1 in the classes 22-24 and 27, Figs. 4 and 5. So, from a symmetrical point eif view class 13 equals class 16, and 12 and 14 are equal to 15 and 17. Having observed this we should be able tei use the frequencies in the weight classes 13 to 21 among the coins in chain 1 and the frequencies in the classes 8 to 15 among the round coins of chains 10 and 11 tei reconstruct the symmetry once intended. When you start to calculate this mean distribution you will soon discover that it is probably a distribution consisting of 96 units in 11 classes of the following frequency proportions: 1 : 4 : 7 : 12:16:16:16:12:7:4:1. This means that each of the three central classes contains 1/6 of the material or half of it together. The third fourth of the coins are feiund in the two classes next to the three central ones, and the last fourth in the six extreme classes, Fig. 10. Tested with a ;f-test for n— 1 = 8 degrees of freedom (the expected values of the extreme classes are too low to fit the test) the chance that the sample was drawn from a peipulation with the suggested proportions is greater than 99.9995%. Obviously this is a reconstruction of a possible ideal which was never executed
Fornvännen 87 (1992)
in detail. In Roman metrology, however, similar 96 part distributiems can be feiund (Herschend in press). What Olof had in mind In the mind of Olof Skötkonung practical economy and metrology mingled with the theoretical ideals of these subjects.
X Chain 10 & 11 dass8-14
Cha-olX BS 15 - 21
Fi
.
12/15
13/18
14/17
Fig. 9. A comparison between the distribution of the seven lower classes of chains 10 and 11 (Fig. 7 blank bars) and their counterparts in chain 1 between 15 and 21 units (Fig. 7 hatched bars). - En jämförelse mellan fördelningen i de sju lägsta viktklasserna i kedja 10 och 11 (Fig. 7 ofyllda staplar) och deras symmetriska motsvarighet i kedja 1 nämligen klasserna 15 till 21 (Fig. 7 skrafferade staplar).
What Olof had in mind
27
Fig. 10. The ideal 96-unit distribution. - Den ideala 96-enhetsfördelningen.
To approach the economic side let us assume that the coins in chain 1 show the coinage which circulated on the Sigtuna märket in the late 990's. Moreover, let us take the symmetry for granted and compare the ideal with the observed practice, Fig. 11. Twei interpretations of the diagram should be emphasized. Either the coinage was designed from the very beginning to be slightly askew, or the symmetrical ideal was abandoned after a few years. The asymmetry is certainly neit prominent and cannot be compared with the tip of scales established by the låter coinage. The dividing line should therefore run between chain I and chains 10 and 11, and not between the ideal and the initial coinage. Thus one should understand the few heavy coins in chain 1, in the mean weighing 24 units instead of 16, as a Iure for the merchants to increase their turnover and thereby be in a better position to obtain the few heavy coins, an increase from which even the King would profit. To compensate for the heavy coins without losing silver the coinage must contain some light ones, and instead of making a few very light, but eibviously counterfeit coins, the balance is created mainly by an over-production of coins in the classes of 11 and 12 units. In the diagram the coins below the ideal amounts to 190 units and those abeive to 189 units. That too is an indication of the King's will to accelerate coin circulation rather than falsify the coin weights. Knowing what happened a few years after the introduction of O l o f s denier, when he
started to circulate coins like those in chains 10 and 11, we can be relatively confident that Olof had at least two consecutive steps in mind, when he decided to engage in minting. First he took the expensive step toward the creation of confidence in, and a need for, money. For this reason he started by introducing a coin which was both a reasonably fair currency and a produet of some interest to those who profit from inter märket exchange. He then introduced the more lucrative coinage with emly a few coins to satisfy those who sought to match or outdo the original real value definition of the coin. The difference between the heavy parts of the distributions can be described in a more precise terms as follows: In chain 1 the average number of coins more than 5 weight units heavier than the norm is 3.4%. In chains 10 and 11 the corresponding percentage is 1.6%. According to the Poisson distribution this means that the chance of obtaining more than two of these heavy coins in a sample of 100 coins in chains 10 and 11 is 21.6%, while the chance of finding more than four such coins in a corresponding sample of chain 1 ceiins is 29.4%. O n e might say that to small dealers a coinage like that of chains 10 and 11 is of little interest from the point of view of real value. The King and the tradesmen with a large turnover benefit from this in a way we would today find unfair.
Fig. I L A comparison between the expected symmelrical distribution (blank bars) anel lhe observed (hatched bars) distribution of the coins in chain 1. En jämförelse inom mynten i kedja 1 mellan den förväntade symmetriska fördelningen (ofyllda staplar) och den observerade (skrafferade staplar).
Fornvännen 87 (1992)
28
1536
F. Herschend equal tei its weight and silver content. After a few years he added the value of his märket to the equation. This allowed him in principle to reduce the weight of the coin without (hanging its value, thereby making the coin less relevant to the real weight economy outside the märket. In this paper the development has been described as a simple transition from one principle to another. Generally speaking this may have been true, bul in reality it is difficult to convince people that the value of the märket should be reflected in the weight of the coins used there: Olof probably needed privilege as well as political power both to drive home his message, and to balance the two economics of real and nominal weight. Even though the transition between the two systems could well have been abrupt, it is also possible that the King made several vain attempts to introduce the lighter coins, and even when he had succeeded, there might freim time to time have been a need to strike some obviously heavy coins, e.g. the square ones, in order to satisfy the need for ceiins with a high real value. The reconstruction of the metrology in the Mälar Valley is advanced by the analysis, which established thal trade weight, the Gotlandic bracelets, and coin weight, tbc Sigtuna coinage, refer naturally to the same mark. The most interesting fact beside the mutual suppeirt of the two materials, is the definition of the coin weight as equal tei 16 smaller units. This means that the mark consists of 1,536 uuils. This small unit, the common denominalen (c, d. in Fig. 12) ofall weights, is difficult to name, but was probably a trade weight. The question is now whether the drachma weight in the Gotlandic bracelets, i.e. 1/64 eif a mark, (Herschend 1987 p. 184 f) or the coin weight in Sigtuna, is the more natural fraction of the metrology, Guided by Medieval knowledge of the mark division system in Svealand, (e.g. Rasmusson 1966 col. 438) one would favour the coin weight, 1/96, in this respect. It is a quarter of an örtug while- the drachma is one eighth of an öre and not a natural li at tion of an örtug. So far the drachma is known only from Gotland and there it is a weighl connected with adjusted jewellery. Thus the recon-
26.06
2.172
Fig. 12. Mark division and weight in Svealand c. 1000 A. D. - Markdelningssyslemet och dess vikter i Svealand ca 1000 e.Kr.
Some traits in the coin weights of the late coinage suggest that there was even then from time to time a positive ring to the King's opinions in the fields of economic decency and necessity. O n e such trait is the lack of coins in the classes below eight units. There are seven coins of both forms and all three chains in that dass. Compared to the distribution of the coin weights of chain 1 this is odd, since from that currency yeiu would expect the extreme to be- a coin in the dass of say 5 units. It seems in other words as if the King, perhaps for some moral reasons, was opposed tei the circulation of coins of less than half of the stipulated weight. Tinning to the square coins, geeimetry teils us that a round coin of a give-u diameter weighs about 6 3 % of a square ceiin with sides equalliug the diameter of the former. Thus there are five coins among lhesquare ones that would have fallen below the dass of 8 units had they been stamped out round. Even this is a hint tbat there was a le-si at the 8-unit levd, which all coins must pass. To summarize, one might say thal Olof stalled by recognizing that the value ol a coin was
Fomvännen 87 (1992)
What Olof had in mind struttion of the metrology in Svealand c. 1000 A.D. should resemble Fig. 12. It seems evident that Olof Skötkonung cheise to use a natural fraction of the mark weight to defme bis coin. When after centuries of precise metrological practice in connection with bulliem you make your first local coin, this is your only recourse, if your purpose is to create confidence in the currency. A denier silver coin is the only respectable starting point. The last point to be stressed in an artide about the weight of the Sigtuna coinage is the definition of the ceiin as a denier, and many readers will feel that they have read that once or twice somewhere before. It might well have been in Hildebrand 1887 or in Thordeman 1936. The mean weights which made u p the point of departure for Hildebrand or Thordeman were not correct and their insights in the complex Sigtuna numismatics not those of today, but in their eiverall conceptual understanding of the rough mean weights, guided by common sense, they were neverthdess right in their interpretation. In a confused heap of coin types and weights they used Ockhanfs razor and favoured a simple solution where it could be found—arguing that a ceiin weighing a little more than 2 gramnies in Middle Sweden c. 1000 A.D. is probably a denier. Thordem a n ^ approach was strongly critidzed by Malmer 1965. Having showed the complexity of the coinage Malmer c o n d u d e s : To deduce only one delinile coin asscssmenl system from a material so lacking in uniformity is of course impossible. On the contrary, the norms were conslanlly shilling and lhe boundary belween them i-. moreover in some cases floaling. (Translated from Malmer 1965 p. 29.) Altheiugh " d e d u c e d " is hardly the right word to describe Theirdeman's procedure. Beith authors were in my opinion right, as well as wrong. Thordeman pointed out a probable ideal which may very well have existed. The coin weights he used did not represent the coinage as we see it teiday, and bis absolute numbers were wrong as were Hildebrand's, but they suffice to ceimprehend an ideal.
29
Even Malmer's opinion is reasonable. It is in the nature of things that there is little tei deduce in these matters d u e to the uncertainty of the premisses, but there is reason to believe that the ceiins do mirror several norms. Yet this does not mean that these norms need to be defined in different metrological systems. On the contrary, in this casestudy 20 different norms belonging tei the same system (the weight classes 8 to 27) are used to describe the complexity eif the weights, and the initial ideal (the denier) remains. From a methodological point of view it is interesting to see that the three major die chains in the early Sigtuna coinage, established in Malmer 1989, have worked in principle as Hildebrand's and Thordeman's rough mean weights. Both the mean weight and the chains are ways of cutting through complex patterns which cannot readily be graspe-cl. Such a cut does not disdose the intricate complexity, but it may give a d u e to a general structure not otherwise found. References Blackbum, M. 1985. English dies used in Scandinavian imitative coinage. Hikuin II, Hojbjerg. Herschend, F. 1987. Metrological problems. Tor 21, Uppsala. — 1989. Vikings following Cresham's Law. In: T. Larsson 8c H. Lundmark (cd.) Approaches lo Swedish Prehistory. BAR Int. Ser. 500. Oxford. — in press. A case-sludy in metrology—the Szikane s Hoard. Tor 23, Uppsala. Hildebrand, H. 1887. Sveriges in\iil under medeltiden. Stockholm. Kyhlberg, O. 1973. Vikllod. Birka. B. Ambrosiani 8c B. Arrhenius (ed.) Svarta jordens /uiiiDiiiiiiråde. Arkeologiska undersökningar 1970—71. Riksantikvarieämbetets rapport C 1. Stockholm. — 1980. Vikt och värde. Stockholm. — 1982. Vikter och värderingar—en genmäle. Fornvännen. Sioe kholm. Lindqvist, S. 1956. Frän Upplands forntid. Kort vägledning genom Uppsala universitets museum för nordiska fornsaker. 3. rev. ed. Uppsala. Lundström, L. 1973. Bitsilver och betalningsringar. Studier i svenska depäfynd påträffade mellan 1900 och 1970. Stockholm. Malmer, B. 1965. Olof Skötkonungs mynt och andra Ethelred-imitationer. KVHAA Antikvariskt arkiv 27, Stockholm. — 1989. The Sigtuna coinage. Stockholm. Melcalf, M. 1987. Hexham and Cuerdale: Iwo notes Förmannen 87 (1992)
30
F. Herschend Steuer, H. 1984. Feinwagen und Gewichte als Quellen zur Handdgeschichte des Ostseeraumes. In: H. Jahnkuhn, K. Schietzel, H. Reichstein (ed.) Archäologische und naturwissenschaftliche Untersuchungen an ländlichen und fruhstädtischen Siedlungen im deulschen Kustengebiet vom 5. Jahrh. v. Chr. bis zum 11. Jahrh. n. Chr. Bd 2, Handelsplätze des friihen Mittelalters. Bonn. — 1987. Gewichtsgeldwirtschaften in friihgeschichtlichen Europa. In: K. Diiwel et al. (ed.) Untersuchungen zu Handel und Verkehr von der vor- und friihgeschichtlichen Zeit in der Mittel- und Nordeuropa. Teil IV. Der karolinger und Wikingerzeit. Göttingen. Thordeman, B. 1936. Sveriges meddtidsmynt. /Vordisk kultur 29. Stockholm.
on metrology. In: D. M. Metcalf (ed.) Coinage in ninth-century Northumbria. BAR British Series 180. Oxford. Petersson, B. 1969. Anglo-Saxon currency. King Edgar 's reform to the Norman conquest. Lund. Saers, J. 1982. En kritik av Kyhlbergs viktsats-analyser. Fornvännen. Stockholm. Skaare, K. 1976. Coins and coinage in Viking Age Norway. The establishment of a national coinage in Norway in the XI century with a survey of the preceeding currency history. Oslo. Sperber, E. 1988. How accurate was Viking Age weighing in Sweden? Fornvännen. Stockholm. — 1989 a. The find from Bandlunde, Gotland: 150 weights belonging to an Islamic weighl system. iMborativ arkeologi 3/1988. Stockholm. — 1989 b. The weights found at the Viking Age site of Paviken, a metrological study. Fornvännen. Stockholm.
Sammanfattning
Avsikten med fallstudien är att söka förstå viktfördelningen hos de första Sigtunamynten, sådana de framstår i Malmer 1989. Till grund för ett sådant försök att utnyttja en materialpublikatiem ligger en teoretisk uppfattning om Sigtuna som marknad och om myntens funktion. Den förhistoriska ekonomin leder till ett silveröverskott i Mälardalen. I en outvecklad ekonomi går det inte att förbruka det silver och guld, som man av och till kan ha turen att komma över vid utfärd. För människor som deltar i mer utvecklade ekonomier kan det vara intressant att handla sig till detta silveröverskott på lokala marknadsplatser genom att importera varor dit. Lika intressant är det dock för en begynnande kungamakt att organisera en sådan marknad och kanalisera importen till denna mot att de stim begagnar sig av marknaden betalar en form av avgift. Även kungens syfte är att få tag i en del av det silver som befinner sig som inaktiv förmögenhet hos regionens bondefamiljer. Kungens mål måste vara att få marknaden att fungera med övervärderade mynt. Men dessa är knappast naturliga i en region som Fornvännen 87 (1992) finner det naturligt att använda sig av metallvikt i sina värdemetalltransaktioner. Därför är det sannolikt att mynten skildrar övergången, eller en del av övergången, från metallviktekonomi till penningekonomi med övervärderade mynt. Då är det också naturligt om de första mynten har en god anknytning till det existerande metrologiska systemet. Mynten i den klassiska seriens kedja 1 speglar bäst de mynt som cirkulerade under de första åren. De avslöjar sig som en i huvudsak symmetrisk fördelning i 11 viktklasser (11 t. o. m. 20) relaterade till en mark om ca 208,6 gram. Det mest frekventa myntet finns i fördelningens centrala klass och dess vikt är lika med 16/1536 eller 1/96 av denna mark dvs. en denar. Myntningens kvalité utvecklas i och med mynten i den nya serien. Det betyder förmtidligen att marknaden börjar fungera och att rationell myntpreiduktion snarare än metrologiskt lätt genomskådliga mynt blir viktig. Viktfördelningen ändrar sig nu radikalt. Det vanligaste myntet kommer att väga 13 enheter i stället för 16 och i stället för att prägla en symmetrisk fördelning i 11 klasser präglar
What Olof had in mind man huvudsakligen de 7 lägsta viktklasserna (8 t . o . m . 14). Eftersom man förmodligen fortfarande rör sig med ett mynt, som nominellt är en denar, ser man troligtvis en begynnande
31
tendens till övervärdering av myntens värde, Detta är ett uttryck för kungens makt och Sigtunamarknadens etablering.
Fornvännen 87 (1992)