Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors

The Bag from Roswinkel, Netherlands



The following article aims to collect all known data on a bag found in Roswinkel, the Netherlands, dating to the end of the 9th century. It is most likely the most complete find of a Carolingian period bag from continental Western Europe to date. Despite the uniqueness of the find, the available data are fragmentarily scattered across the literature and some details are not known to the public at all. Significantly more attention has been paid in the archaeological academia to the coins of the hoard, while the bag stands in seclusion.

Our efforts for proper publication will be complemented by detailed scaled photos kindly sent to us from the Drents Museum in Assen, the Netherlands. For the first time ever, the back side of the find will also be published. The text includes a scheme mapping the approximate construction process and a practical reconstruction of the organic component, which can serve reenactors and others interested in the practical everyday life of the Early Middle Ages. All works are created with the consent of the above-mentioned museum and its staff.

Location of Roswinkel on the map of Europe.

Previous publication

The organic part of the hoard (bag and box) was first published not long after its discovery, when it was published ina colour illustration by Pleyte in 1883 (Pleyte 1883: Pl. XXIV). The illustration also included approximate scales. The digitized version of this old and inaccessible source was officially published around 2022, so the vast majority of interested parties used the black-and-white redrawing without scale published in an article by Gräslund (1984: Abb. 16.2). The Dutch expert on leather goods Goubitz (2009: 39) also adopted an identical redraw. Haertle only mentioned the existence of a leather bag and a wooden box in his book (Haertle 1997: 158). The year 2021 brought a new wind to the bag study, when the bag first appeared in the temporary exhibition of the Drents Museum. On that occasion, two popular articles were published that depicted the current form of the find – one article was published in press (Huisman 2021), the other was published online (Drents Museum 2021). In Huisman’s article, a scaled photograph appeared for the first time.

Incomparably more attention was paid to the coins found inside the bag from the beginning of the discovery. It would be pointless to list all the numismatists who included coins from Roswinkel in their work. Let’s name at least the most important ones. Recognized catalogs that identify coin types and their numbers include Coupland’s article (Coupland 2011: 218-9), Gelder’s article (Gelder 1961: 40) and Haertle’s book (Haertle 1997: 158-161, 687-704, Taf. 40-44, Cat. no. 75). A golden solidus from the hoard was included to the Grierson’s study (Grierson 1951: 9). Let us add that the hoard is labeled “Roswinkel 1870” in the numismatic works to distinguish it from the 1877 find from the same location, which contained two coins (Haertle 1997: 1021).

Archival illustration of the bag from Roswinkel.

Circumstances of the find, place of storage, dating

The discovery was made in May 1870 during peat extraction between the villages of Emmen and Roswinkel. At that time, peat digger J. Kremer and his colleagues came across a hoard of 144-145 Carolingian coins, some of which were stored inside a leather bag, some of which were located in close proximity to it (Haertle 1997: 158; Huisman 2021: 16). The hoard also includes a decorated copper alloy sheet, rolled into a tube (length 7.4 cm, diameter 1-1.2 cm). The bag was supposed to be rolled up and closed with a leather cord (Huisman 2021: 17). One of the coins is gold, the others are silver. At least one silver coin was kept by priest Stratingh from Roswinkel and the rest of the hoard went to the museum for inspection. After cleaning, the set of coins weighed 0.227 kg. The bag, the copper alloy tube and at least some of the coins are now housed in the collections of the Drents Museum in Assen, Netherlands under the inventory number 1870/V. The inventory number of the bag is 1870/V.1. Next to the bag, a small wooden box with a sliding lid and some coins inside was located. The box is stored in the same museum with the inventory number 1870/V.2. The inventory number of the rolled tube is 1870/V.3.

Photo of the Roswinkel bag inventory card.
Source: Drents Museum collections, Assen.

In terms of dating, there are no radiocarbon or dendrochronological analyzes of the organic objects. The dating thus necessarily follows from the attached coins, which were minted during the reign of Charlemagne, Louis I the Pious, Pippin II of Aquitaine, Lothar I, Charles II the Bald and Louis II or Louis III. The oldest coin dates from 793/4-812, the youngest from 877-879/879-882 (Haertle 1997: 159-160). There is a tendency among numismatists to date hoard including the one under discussion to the period immediately following the mintage of the youngest coin. This is the case of Coupland and Grierson, who date the hoard to roughly 880 or between 877-885 (Coupland 2006: 249; 2011: 218; Grierson 1951: 10). The author contacted Simon Coupland with a question on what basis one can infer the correctness of such an approach. Let’s hear the answer in full:

Essentially the Carolingian rulers introduced new types on a regular basis, especially early in the period, and coin hoards show clearly that:

  1. during the period of empire the earlier coins were very successfully removed from circulation;
  2. even after that, earlier coins disappeared through natural wastage after only a few decades.

Hoards from within the empire can therefore be dated with reasonable accuracy: the absence of later issues indicates that the hoard was concealed before their introduction.

These are valid arguments that undoubtedly reflect a part of reality. However, the Roswinkel hoard clearly shows that a number of coins were in circulation for more than a quarter of a century, so the success of removing coins from circulation was certainly not 100%. The absence of issues of new coins of Charles III the Far, Carloman of Bavaria or Odo of France is a convincing point. Along with this, it is necessary to draw attention to Haertle’s comment that the coins were deposited in the ground in a very good condition, which indicates that they could not have been used for a long time (Haertle 1997: 158). Despite plausible reasoning, the author leans towards a more secure dating of the deposition of the hoard in the last quarter of the 9th century after 877.

The placing of hoard in the ground in the 9th century Frisia is traditionally interpreted as an effort to hide valuables from Scandinavian raiders (Armstrong 1998). Although the specific reasons for the Roswinkel hoard will never be known, Coupland has shown that multiple factors may have been at play, such as an increase in the money supply and higher taxation of the population (Coupland 2006: 243). He also pointed out some interesting facts: when the Dorestad mint disappeared around 855, the number of Frisian hoard rapidly decreased and the hoard began to consist of a wide variety of coin issues (Coupland 2006: 262; 2010). The find from Roswinkel is an illustrative example of this.

Bag, box and some coins from Roswinkel.
Source: Drents Museum collections, Assen.

Description of bag and box

The bag stands out for its unexpectedly large dimensions, which make it difficult to call it a “wallet”. A more appropriate name is “bag”. According to the inventory card, the bag is 36 cm long and 16 cm wide. Judging from the attached photos, the current outer dimensions are closer to 32 × 14 cm without the cord, so it is possible that the leather has shrunk. Considering that the bag contained a large amount of silver, these are not excessive parameters.

In terms of construction, the bag consists of five parts and a cord. The back half is two-part, almost complete and has a lid and the cord. The front half, on the other hand, is three-part, partially damaged, and its sewing to the back half created one long pocket about 28 cm long. The material for production was about 0.1 cm thin tanned deer skin (reviewed by Dr. Carol van Driel at the turn of 2021/2022). It is obvious that the parts of the individual halves were sewn first and then the front and back halves together. For the front half, the middle part is covered by both adjacent parts; in the back half, the longer part overlaps the part that co-creates the lid. The parts overlap in length by approximately 0.8 cm.

It must be added that the current form is a reconstruction, as the original stitches have not been preserved and the stitches present today are a not very attractive and relatively old conservation intervention. This fact explains the sloppy matching and stitching of the two halves as well as the finishing knots at the edges of the bag. It can be assumed that the bag was originally sewn with a thin linen thread with a single-row straight saddle stitch. The holes have a spacing of 0.3-0.5 cm and were made with an awl with a circular cross-section.

Photos of both sides of the bag from Roswinkel.
Source: Drents Museum collections, Assen.

An imperfectly centered triangular lid with a rounded top is placed on top of the reverse half, which still shows coin impressions. Two holes are made in the top, through which a cord with a width of at least 0.5 cm is threaded and then knotted. This is inserted from the inside so that the knot is inside and not visible from the outside. The cord is broken; in Pleyte’s drawing it is complete, which may be a reconstruction, but it cannot be ruled out that the rupture occurred only during storage in the museum. The ends were tied together after breaking. The length beyond the lid of the bag after tying is at least 12 cm, before breaking it was at least 15 cm.

Detail of the cord. Source: Drents Museum collections, Assen.

On two parts of the inner half, there are visible traces of the now-absent stitched decoration, which took on a rhombic shape with a superstructure in the shape of a regular cross. The decor was sewn with a smaller stitch with shorter spacing than is the case with the parts of the bag. A close reading of the prints and the layout of the holes showed that each part had been decorated before being stitched together. The decoration was therefore not made up of a single openwork leather appliqué. The decor of the lower part almost certainly reached below the level of the neighbouring part. The decoration was more likely achieved using split bands (six and five) rather than single cut pieces. The pieces were sewn together in such a way that the decorative bands met to form a pattern, which was not done perfectly, as the center lines illustrate.

Approximate construction diagram of the bag from Roswinkel. Author: Diego Flores Cartes.

The effort to identify the motif and its symbolic meaning is not an easy task, which results in a fascinating, but difficult to verify and potentially wrong theory. The key to reading the motif is the decoration of the central part of the inner half, where there is a pair of vertical beams and a pole extended into a cross. The cross is not found on the adjacent part of the decoration, so it can be assumed that the motif was not symmetrical and interchangeable on both sides. A similar double beam with a cross at the top is a common feature of temple depictions on Frankish coins, including those found in the hoard. However, if the motif was to represent a temple, its base would probably have been rectangular, rather than formed by two more pairs of chamfered beams. The only conceivable close motif is the ship symbol, which was minted by Charlemagne and Louis the Pious in Dorestad and Quentovic and subsequently copied in England and Denmark. The shape of the base is indeed bevelled, but if we assume that the creator worked only with straight strips of leather when saving the material, the result could not have been rounded. In such a case, it would be a schematization: the upper pair of beams would represent the rigging, the middle strip the mast, and the lower pair of beams the hull of the ship, which is also formed by double bands on the coins. As coins with the ship motif do not appear elsewhere in the Frankish Empire, the ship motif is thought to represent the great seamanship of the Frisians (Hodges 1989: 90). Coupland literally says that the ship was a symbol of the two mentioned cities (Coupland 2007: 119). A supporting argument is the fact that the hoard was found only 145 km from Dorestad as the crow flies. If we proceed to theoretical considerations, the symbol of the ship in Frisia may have had connotations of royal power, which was demonstrated on the coins. If we use an early medieval parallel from another part of Europe, the symbol of the ruler can mark collected taxes intended to be handed over to the lord (e.g. Janin 2007: 266–269). This could also be the case with the bag motif from Roswinkel, theoretically stressed by gilding or other accentuation of the leather strips. Should this theory be incorrect, we have at least shown that the closest analogies of the motif are to be found on Frankish coins of the 9th century.

Comparison of the bag motif with the ship motif of the Dorestad mint of Louis I.
Author: Diego Flores Cartes.

An equally impressive part of the hoard is the wooden box, which has been preserved in very good condition. According to the available literature, it is made from one piece of walnut wood (Haertle 1997: 158; Huisman 2021). According to the inventory card, the cuboid body of the box is 9 cm long, 3 cm high and 3 cm wide. A guide notch is created in the upper surface for inserting the lid. On both longer side walls, there are two trios of circular holes (one hole placed above, two holes below), which were used to fasten the straps. The impressions of these straps are still visible and they bifurcate towards the bottom pairs of holes. The purpose of these straps was to hold the lid in place. The box is cracked at the level of the holes. The lid is also preserved and is equipped with a diamond-shaped handle about 2.5 cm long. The cross-section of the lid appears to be plano-convex. There are two small holes of unknown function in the lid, one larger and oval, the other smaller and circular. It would theoretically be possible to store the box in a bag as well.

Box from Roswinkel. Source: Drents Museum collections, Assen.


Appendix 1: All photos provided by Drents Museum.


There is no doubt that the Roswinkel bag is one of the best-preserved pieces of its kind in early medieval Europe. The find is usually compared to folding wallets woven with leather bands (Gräslund 1984: 145; Huisman 2021: 17) found in Birka and Trelleborg (Gräslund 1984: 143-5; Vlasatý 2022). In fact, we have a closer analogy – a fragmentarily preserved bag from Iholm, Denmark (Vlasatý 2020). Only the gilded openwork appliqué, sewn onto a small piece of the bag, survived of the bag, which held at least 475 coins. The dating of the find points to the 1st quarter of the 11th century. A wallet decorated with engraved and embossed lines forming an ornament is also known from Sigtuna and is dated to the years 1030-1050 (Vlasatý 2024). If we were to focus on the regional context, the closest parallel is undoubtedly the partially preserved undyed linen pouch that was part of the Wirdum hoard, dated to around the 840s (Coupland et al. 2021: 130-1). Part of the leather strap is still stuck to the fabric.

Suggested drawn reconstructions of Iholm wallet.
Made by Tomáš Cajthaml.

The one-piece sliding-lid box has numerous analogies that show that structurally analogical pieces have a tradition in European space from at least the early AD years to the Middle Ages. The oldest pieces we will present are from the Germanic Roman Age and Late Antiquity period; representative pieces come from Nydam, Garbølle and Vimose in Denmark and from Dyster in Norway (Rau 2010: 469-470, Taf. 65). A number of boxes used as reliquaries are preserved from the period (Quast 2012: 55, 91). From the period of 8th-11th century we can name finds from Barshalder 22/1936 (Thunmark-Nylén 200: 247), Dublin (Lang – Caulfield 1988: 6, 8, 51-2), Haithabu (Westphal 2006: 85, Taf. 64.5), Moščevaja Balka (Jerusalimskaja 2012: 320-327) and Sigtuna (Sigtuna Museum 2024). Several boxes were found in Novgorod and date to the 13th and 14th centuries (Kolchin 1989: 137, Pl. 129.5, 138.9). A similar box from the 16th century was found in the Norwegian glacial pass at Lendbreen (Pilø 2000).

Boxes are usually wooden (analogical pieces are pine and yew), less often metal. They vary in length from 5-40 cm. Some are decorated on the outside or written with inscriptions. The storage space is usually undivided, but some pieces are divided by compartments. Available finds testify to the storage of coins, needles, jewellery and even candles. Fastening by means of leather straps riveted to the walls of the box is not preserved in all cases; illustrative examples of leather straps are the grave find from Barshalder 22/1936 and the find from Lendbreen.

Box from grave 22/1936 from Balshalder, Gotland.
Authors: Trotzig 1991: Pl. 7f; Jimi Appel Johansson; Ola Myrin, Historiska museet/SHM.


Professional reenactor and crafter Roman Král (King’s Craft) participated in our evaluation of the find. In the past, Roman created several functional versions of the bag, the dimensions of which were underestimated due to the lack of scale, and the versions served as wallets. For the purposes of this article, Roman has now made a modified prototype of the bag and the box that matches the dimensions and construction procedures. The prototype presented below was made without the gilding of the decorative application.

According to Roman, the bag is not a complicated product and anyone who has basic knowledge of working with leather can handle it. The parts are simply folded over each other, there is no turning and shaping that would require advanced skills. It is a simple utility object that is made up of imperfectly connected parts and whose main purpose is to efficiently transport content of significant value. If the decorative application was gilded, specialized tools (a press) and gold leaf would be needed, which would make the product unusual and above-standard. The total number of hours devoted to production from a sheet of leather is 7-8 hours. Only knives, awls, a pair of needles, linen thread and wax were used.

Roman evaluates the box in a similar way. It is a simple, practical product lacking decoration, which follows the efficient transport of objects. The production of the box took about 4 hours and required a saw, a drill, a chisel, a file and a knife.


The author team would like to wholeheartedly thank Hester Hazenberg and Carole Steenbergen from the Drents Museum in Assen, who willingly answered our questions, did not hesitate and photographed the artifact for the purposes of our article. We express our gratitude to Bart Apeldoorn and Annemies Tamboer, who alerted us to the fact that the bag was on display, shared with us contemporary photos, and thus prompted the creation of the article. We are indebted to Simon Coupland (University of Cambridge) who consulted us on the analogies and coins of the hoard. We would like to thank Jimi Johansson for the opportunity to use his photo. Roman Král (King’s Craft) created an updated model for the article and contributed advice to the description of the original decoration, for which he deserves our great recognition and thanks. We must also mention Diego Flores Cartes and Tomáš Cajthaml (, who are the authors of the diagrams.

We hope you liked reading this article. If you have any question or remark, please contact us or leave a comment below. If you want to learn more and support our work, please, fund our project on PatreonBuymeacoffee or Paypal.


Armstrong, Simon (1998). Carolingian Coin Hoards and the Impact of the Viking Raids in the Ninth Century. In: Numismatic Chronicle 158, 131-164.

Coupland, Simon (2006). Between the devil and the deep blue sea: hoards in ninth-century Frisia. In: Cook, Barrie – Williams, Gareth (eds.). Coinage and History in the North Sea World, c. AD 500-1250, Leiden, 241–266.

Coupland, Simon (2007). Carolingian Coinage and the Vikings Studies on Power and Trade in the 9th Century, Ashgate.

Coupland, Simon (2010). Boom and Bust at 9th-century Dorestad. In: Willemsen, Annemarieke – Kik, Hanneke (eds.). Dorestad in an International Framework, Leiden, 95-103.

Coupland, Simon (2011). A Checklist of Carolingian Coin Hoards 751-987. In: The Numismatic Chronicle 171, 203-256.

Coupland, Simon et al. (2021). A Carolingian coin hoard from Wirdum (Friesland, the Netherlands) and the Dorestad mint. In: Willemsen, Annemarieke – Kik, Hanneke (eds). Dorestad and its Networks. Communities, Contact and Conflict in Early Medieval Europe, Leiden, 119-135.

Drents Museum 2021 = Oude gebruiken en een nieuw geloof. In: [online]. [2024-01-06]. Available at:

Gelder, H. E. van (1961). De Karolingische Muntslag te Duurstede. In: Jaarboek voor Munt- en Penningkunde 48, 15-42.

Goubitz, Olaf (2009). Purses in Pieces: Archaeological Finds of Late Medieval and 16th Century Leather Purses, Pouches, Bags and Cases in the Netherlands, Zwolle.

Gräslund, Anne-Sofie (1984). Beutel und Taschen. In: Arwidsson, Greta (ed.). Birka II:1. Systematische Analysen der Gräberfunde, Stockholm, 141-154.

Grierson, Philip (1951). The Gold Solidus of Louis the Pious and its Imitations. In: Jaarboek voor Munt- en Penningkunde 38, 1-41.

Haertle, C. M. (1997). Karolingische Münzfunde aus dem 9. Jahrhundert, Köln.

Hodges, Richard (1989). Dark Age economics. Origins Of Towns And Trade, A.D. 600-1000, London.

Huisman, Floor (2021). Meer dan munten – nieuw onderzoek naar een oude schat. In: De Spieker – periodiek van de Drents Prehistorische Vereniging 42, 16-18.

Janin, V. L. (2007). Středověký Novgorod v nápisech na březové kůře, Červený Kostelec.

Jerusalimskaja 2012 = Иерусалимская, А. А. (2012). Мощевая Балка. Необычный археологический памятник на Северокавказском шёлковом пути, Санкт-Петербург.

Kolchin, B. A. (1989). Wooden Artefacts from Medieval Novgorod, Oxford.

Lang, J. T. – Caulfield, Debbie (1988). Viking-age decorated wood: a study of its ornament and style, Dublin.

Pilø, Lars (2020). The Hunt for the Lost Mountain Pass. In: Secrets of the Ice [online]. [2024-01-06]. Available at: Available here.

Pleyte, Willem (1883). Nederlandsche Oudheden van de vroegste tijden tot op Karel den Groote, Drenthe, Leiden.

Quast, Dieter (2012). Das merowingerzeitliche Reliquienkästchen aus Ennabeuren. Eine Studie zu den frühmittelalterlichen Reisereliquiaren und Chrismalia, Mainz.

Rau, Andreas (2010). Nydam mose 1-2. Die personengebundenen Gegenstände. Grabungen 1989–1999, Højbjerg.

Sigtuna Museum 2024 = Lock. In: Sigtuna Museum & Art [online]. [2024-01-06]. Available here.

Vlasatý, Tomáš (2020). The Wallet from Iholm, Denmark. In: Project Forlǫg – Reenactment and science [online]. [2024-01-07]. Available at:

Vlasatý, Tomáš (2022). The Wallet from Trelleborg, Denmark. In: Project Forlǫg – Reenactment and science [online]. [2024-01-07]. Available at:

Vlasatý, Tomáš (2024). The Wallet from Sigtuna, Sweden. In: Project Forlǫg – Reenactment and science [online]. [2024-02-14]. Available at:

Thunmark-Nylén, Lena (2000). Die Wikingerzeit Gotlands IV:1–3 : Katalog, Stockholm.

Trotzig, Gustaf (1991). Craftsmanship and function : A study of metal vessels found in Viking Age tombs on the island of Gotland, Sweden, Stockholm.

Westphal, Florian (2006). Die Holzfunde von Haithabu. Die Ausgrabungen in Haithabu 11, Neumünster.

Leave a Reply

Your email address will not be published. Required fields are marked *