The biological and geological record of the transition from the Late Glacial to the Early Holocene is manifested by a dramatic change in vegetation due to climatic warming1. With the increase in temperature vast amounts of buried stagnant ice gradually melted forming water-filled depressions. The Early Holocene landscape was therefore characterized by numerous shallow lakes and ponds in a relatively open birch and pine-dominated woodland2. These formed crucial hunting and fishing grounds for the first Maglemosian people living in Southern Scandinavia.

Throughout the Early Mesolithic barbed bone points were frequently lost in the lakes, presumably in connection with spearfishing. These lakes gradually evolved into bogs where peat accumulated. Fuel shortages, especially during the First and Second World Wars, resulted in industrialized peat exploitation in these bogs, which in turn caused these artefacts to be uncovered and recognized for what they are3,4,5.

The barbed bone points soon became closely associated with the Maglemose culture (c. 11,000–8000 BP). However, despite a long history of research on the typology of these characteristic items6,7,8, their chronological placement has largely been indirectly deduced from stratigraphy and pollen dating4,9. Species identification of bone points has previously been based on such evidence as bone debitage from habitation sites, or the absence of specific skeletal elements in a given faunal assemblage3,10, p. 278,11, which makes the identification of each artefact circumstantial. Based on these types of analyses, the majority of the Maglemosian bone points were thought to have been made from both ribs and long bones of “large ungulates”, translating to aurochs, elk, red- and roe deer (Bos primigenius, Alces alces, Cervus elaphus, and Capreolus capreolus)12.

We analyzed 126 bone points from Southern Scandinavia (Fig. 1 and Supplementary Information, Section 1) using a combination of morphological typology, radiocarbon dating, and proteomic analyses. Using these techniques, we were able to assess the selection of raw materials and date the typological variation. This study is the first attempt to investigate whether a single artefact type can be used as a proxy for both human and animal populations. We demonstrate that such comprehensive analyses can provide information about species dynamics, resource exploitation, human–environment interactions, and perhaps population mobility. Our integrated analyses provide a robust new framework for Maglemose chronology, that extends to changes in lithic manufacture.

Figure 1
figure 1

Overview of the approximate find locations for the 126 barbed bone points from Southern Scandinavia. All bone points found in paleolake systems, now grown into peat bogs. While the selection is finite the spread indicates the approximate Eastern → Western extent of sediments conducive for preservation (for further information on each artefact including species identifications see Supplementary Dataset 1 and Supplementary Figures 1–22 and 29–30) (Map: digital elevation model was produced using open source Copernicus data and information (from the European Union—EU-DEM layers), and then merged with spatial geographical data in QGIS v.3.19 (, followed by a final correction in Adobe Illustrator v. 24.3 (


Bone points

Local bone point typo-chronologies have previously been proposed for two sites in Sweden, i.e. Rönneholms mosse8 and Motala13. However, since the present study comprises material from a much wider geographical region, this fine-scale classification of the bone points might not be representative. Therefore, to reduce bias, we divided the material into two groups; fine-barbed bone points and large-barbed bone points.

Radiocarbon dating

Prior to this study, only five radiocarbon (14C) dates on Danish barbed bone points had been published14,15. With the addition of previously published dates from Scania in Sweden (n = 20)8, unpublished dates from Denmark (n = 7) as well as 21 new 14C measurements acquired specifically for this project, we were able to model the radiocarbon-date distribution of 50 bone points (excluding double dates on the same artefact) (Fig. 2; Table 1; Supplementary Information, section 2; Supplementary Dataset 1). After calibration into cal. years BP in OxCal v.4.316, the artefacts separated into two distinct phases (Fig. 2). The fine-barbed bone points are all confined to the mid-late Preboreal and the beginning of the Boreal (c. 11,200–10,100 cal BP). The larger-barbed bone points are restricted to the end of the Boreal and the beginning of the Atlantic chronozones (9658–8413 cal BP). Between the fine-barbed bone points and the larger-barbed bone points, there is a clear gap in the radiocarbon ages lasting nearly 600 years. In order to explore this hiatus, we summed the radiocarbon dates and performed a Kernel Density Estimation (KDE) simulation to explore periods of activity. This indicates temporal and spatial morphological patterns and confirms the age correlated distributions of the two types of bone points in both regions (Fig. 2).

Figure 2
figure 2

Radiocarbon dates from bone points showing the separation of the two types. (a) A Bayesian model assuming two phases performed on 52 bone points. Double dates on the same artefact marked with *. Carbon distributions colored to denote finding the place or phase boundaries, (b) Summed radiocarbon dates of bone points (shown in light grey), KDE to visualize activity and hiatus (light blue), boundaries marked in dark grey, (c) Summed radiocarbon dates from habitation in Denmark (see Supplementary Figure 23) (shown in light grey), KDE to visualize activity and hiatus (light blue), boundaries marked in dark grey, (d) biozones (Preboreal onset after Jessen et al.1)(e) the two bone point complexes separated by hiatus (see Table 1 and Supplementary Dataset 1).

Table 1 List of 52 radiocarbon dates and stable isotope data used in this study.

Published and unpublished radiocarbon dates from the faunal remains of Eurasian elk (n = 73), red deer (n = 33), and bovines (aurochs + bison Bison bonasus, n = 73) reveal reduced frequencies in the faunal remains of these species around c. 10,000 cal. BP in Denmark (see Supplementary Figure 24; Supplementary Information, Section 2; Supplementary Dataset 2). However, no equivalent decline in the dates from Southern Sweden has been observed. If this decline in radiocarbon-dated remains from Eastern Denmark is representative of a decline in the animal populations this would have implications for the availability of raw material for the manufacture of bone points. The absence of bone points also corresponds to a partial absence of dated habitation sites in Eastern Denmark (not including the island of Bornholm) and Jutland (Fig. 2). However, it is during this same gap in the Danish record, that most of the dated habitation sites on the west coast of Sweden occur, as well as on the island of Bornholm (see Supplementary Figures 23 and 24; Supplementary Information, section 2; Supplementary Dataset 3).

Protein analysis

One-hundred and twenty barbed bone points were analyzed by ZooMS17 (see Supplementary Information, section 3, Supplementary Dataset 4). They turned out to derive from three groups of mammals: 74 from cervids, 43 from bovines and three from brown bear (Ursus arctos) (Fig. 3).

Figure 3
figure 3

Species identifications of bone points. (a) Histogram of 120 summed protein mass spectrometry identifications separated by radiocarbon hiatus (see Supplementary Dataset 4). Each column refers to barbed bone points made of respectively long bone and rib. Colours refer to stylized animal portraits. Gradient colours (blue vs. orange) indicate either red deer or elk, (b) selected bone points of each class.

Similarly to aurochs and bison18, red deer and elk could not be distinguished using previously published markers, e.g., from Welker et al.17. However, through mining of published whole-genome data, we were able to construct more complete sequences for the species revealing five single amino acid polymorphisms (SAPs) between red deer and elk resulting in five potential tryptic peptides (biomarkers) (Supplementary Table 1; Supplementary Information, section 4). In three of these biomarkers, one or both sequences also matched (100%) with environmental bacterial sequences as identified using BLASTp and were discarded (Supplementary Information, section 4). LC–MS/MS analyses of four reference samples (two red deer and two elks) confirmed that the two remaining biomarkers can be used to discriminate between the two species (Supplementary Information, section 4). In the MALDI spectra, the red deer peptide m/z 2216 (GETGPAGRPGEVGPPGPPGPAGEK, peptide COL1A1T66/67, position 910-934) results from a missed cleavage because the A → P substitution reduces the efficiency of tryptic cleavage at the adjacent arginine19. As m/z 2216 is also present in aurochs we can only use this marker to discriminate red deer once aurochs has been discounted. Unfortunately, masses corresponding to the equivalent sequence in elk were not detected in any of the MALDI spectra, so we can not confirm that samples lacking the m/z 2216 are elk using ZooMS. Of the 74 bone points previously identified as red deer/elk, 22 were reassigned to red deer. Two Preboreal bone points analyzed by LC–MS/MS (VHM13821 and A37811) had diagnostic peptides for elk, while two from the Boreal period (A40894 and A42422) were identified as red deer, confirming the earlier MALDI-MS based identification (Supplementary Table 2).


Radiocarbon dates from Denmark and Scania (southern Sweden) revealed a hiatus starting at c. 10,300 cal BP, lasting nearly 600 years that separates the Maglemose period into two complexes represented by bone points of a markedly different form. Protein analysis revealed that the fine-barbed points (i.e. Preboreal and Early Boreal) were also materially different from the larger-barbed bone points (i.e. Late Boreal).

Fine-barbed points of the Early Maglemose Complex (n = 45) were predominantly manufactured from cervid long bones (see Fig. 3) despite evidence for the presence of other large ungulates—aurochs, bison, reindeer (Rangifer tarandus), and wild horse (Equus ferus)—in Southern Scandinavia at this time20,21,22. This overrepresentation of cervids may be a reflection of the species abundance at the time but more likely represent a deliberate choice by Early Maglemosian hunter-gatherers. Seven of the 45 were identified as red deer based on our novel biomarker. One of the seven was also radiocarbon dated (Øresø Mølle, A52426, 9481 ± 47) and represents the earliest occurrence of red deer in Denmark.

The bone point morphology of the Late Maglemose Complex changes to a larger and more varied barb design. The number of bone points made of bovine ribs increases substantially (from 6 to 93%) in this large-barbed (n = 75) assemblage (Fig. 3) and those of brown bear appears for the first time. All confirmed bison remains from Denmark and Southern Sweden have been radiocarbon dated to no later than the Preboreal20,23. This seemingly leaves aurochs as the sole persisting bovine species and assumed source of bone points produced after the Preboreal. The bone points made from the aurochs ribs represent the very last remnants of this species before its disappearance on the Danish islands c. 7000 cal BP24,25. The selection of aurochs rather than cervid ribs probably reflects its preferable mechanical properties (Supplementary Information 5). It could, therefore, be argued that the selection of the raw materials by Maglemosian hunter-gatherers was based upon practical, rather than for example spiritual, considerations26.

The majority of the large-barbed bone points date to the Maglemose culture, while seven dates extend into the beginning of the Kongemose culture (c. 8500–7400 cal BP), after which simple bone points without barbs seem to have been preferred (at least in Central Scania)27,28. The transition from large-barbed bone points to simple bone points seems to be synchronous with the arrival of trapezoid lithic armaments29,30,31. This change in material culture may reflect a change of economy and seasonal rounds of the local population: coastal and inland groups merging within Southern Scandinavia as sea levels rose; seasonal spearfishing in the lakes losing its importance by communities that relocated their demographic centres to the coasts during the Kongemose32. However, the transition to simple bone points and Blak type trapezoid armaments might also represent the arrival of new migrants, whose new technologies ended the microlithic tradition completely.

Interestingly, the hiatus of dated barbed bone points is also evident in the radiocarbon dates from the classic Maglemosian habitation sites in Eastern Denmark, in effect dividing them into two periods of occupation (see Fig. 2 and Supplementary Figure 23, Supplementary Information, section 2; Supplementary Dataset 2). Further, a similar gap in radiocarbon dates of faunal remains further indicates a decline in population size in Eastern Denmark, but interestingly not in Scania (apart from Rönneholms mosse) (see Supplementary Figure 24, Supplementary Information, section 2; and Supplementary Dataset 3). The δ13C values obtained from the bone points indicate vegetational change over time from an open environment to a more closed setting (see Supplementary Information, sections 6 and 8). This is unlikely to represent a behavioural change in the animals, but instead probably reflects a landscape characterized by denser forests.

The gap in radiocarbon dates during the Early Holocene might be interpreted as evidence for a decline in human habitation in Eastern Denmark. There is only weak evidence for human presence during the ‘bone point hiatus’. The 14C dates from the two settlement sites of Draved and Klosterlund in Jutland fall before and partly during the hiatus33,34. However, the radiocarbon dates from these sites are conventional and from the very early days of radiocarbon dating when samples were not processed to remove secondary humic acids; a factor that can often result in misleadingly young dates35. Five AMS dates from uniserial bone harpoon points date to the hiatus in Southern Scandinavia. These are from Tunebjerg Øst (9050 ± 40 14C BP) and Trunderup Mose (8845 ± 60 14C BP), both sites on Funen36, from Rönneholms mosse in Scania (8610 ± 90 14C BP)37 and from Vallensgård Mose on the present-day island of Bornholm in the Baltic Sea (9250 ± 60 14C BP, 8875 ± 65 14C BP)38, which would have been connected to Continental Europe at this time.

When large-barbed bone points appear following the hiatus (at c. 9650 cal BP), they show markedly different morphological traits, from their predecessors. They appear once the former major lakes fill and deepen again. Most of the smaller ponds would probably at this time, have grown into fens. Until recently, it was believed that the new bone point morphology appeared before the first indications of the pressure blade lithic industry in Denmark at c. 9000 cal BP39. This industry is characterized by small regular blades, created by applying pressure rather than direct percussion. Blades of this type were utilized as cutting-edge inserts in slotted bone points. However, based on a previously unpublished radiocarbon date obtained from a slotted bone implement (Clarks Type B1 or B2) from Fugle Å, near Ulkestrup Lyng in Store Åmose (AAR-11949, 8360 ± 55 14C BP) (see Fig. 2 and Supplementary Figure 29); pressure flaking is now contemporary with the emergence of large-barbed bone points.

The reason for this gap in the radiocarbon record which seemingly separates two cultural traditions is currently unknown. Several factors could have caused a small and confined human population to almost disappear from the archaeological record; i.e. epidemics, warfare, changes in subsistence strategy, climate, and migration. However, most of these factors are difficult to tie to the disappearance of bone points in the Southern Scandinavian lakes. Several local studies have revealed climatic fluctuations in the Northern Hemisphere during the Early Holocene including lower temperatures and increased precipitation40,41,42,43,44,45,46,47. To our knowledge, no evidence of a similar cool interval or increase in precipitation between c. 10,300 to 10,000 cal BP has been reported for Southern Scandinavia. This may be due to a lack of high-resolution studies of sediments from this period coupled with a sampling bias, further complicated by contemporaneous erosion events48,49,50.

High-resolution studies coupled with direct radiocarbon dating of sediment cores conducted in Southern Sweden, however, do not show an increase in precipitation in lake levels51, 52 at that time. Rather, they indicate dry climate and the lowest water levels of the entire Holocene during the early Boreal51,52,53. The northern expanse of hazel (Corylus avellana) at this time1 is also believed to have been facilitated by the markedly lowered water levels in the lakes and fens20. Pollen analyses of the Åmose basin on Zealand and Rönneholms mosse in Scania also show several marked water level changes that occurred during the Early Holocene48,49, although these are not correlated by direct radiocarbon dating of the sediments. During the transition from the Preboreal to the Boreal, a brief, but significant lowering of the water levels took place, which was shortly followed by renewed transgression54. In Åmosen, this regression led to erosion of the littoral zone and the redeposition of sediments, basically removing most Preboreal riparian sites and sediments associated with this period50. As in Scania, the lowering of the water level was also followed by a transgression48. The reason for these water level fluctuations is difficult to determine. One factor during the Early Preboreal could be the melting of stagnant ice, whereby glacier ice melted in situ. Temperature oscillations may also account for some of these water-level changes, with dryer climatic conditions resulting in increased evaporation55. “The general notion of a particularly dry climate during this period also seems to be reflected in sediment cores from Eastern Denmark, Southern Sweden and Western Germany, which all contain increased levels of charcoal, argued to have been caused by wildfires48, 56,57. Similarly, Jørgensen observed pollen from the shrub Ephedra sp. in Åmosen at the interphase between the Preboreal and the Boreal periods. Ephedra is known to thrive in arid environments, indicating that Åmosen might have been arid at this time period, although he also argues that it could have been transported by the wind48. If the climate did indeed force hunter-gatherers to relocate and abandon their old activity areas, these actions might be reflected in radiocarbon dates elsewhere in the form of potential bust and boom cycles and a rapid diffusion of technologies.

Interruption in 14C dates or drastic changes in the archaeology elsewhere in Europe is also centred around 10,000 cal. BP. In Eastern Fennoscandia, a c. 200-year long gap seems to occur shortly after 10,300 cal. BP, which has been interpreted as a decline in the local population as a result of an abrupt climate event58. Radiocarbon dates of human skeletal remains in Central Germany also reveal a similar gap, indicating either an absence of humans in the period or potential preservation or sample bias59. Interestingly, in North-western Europe this period marks a radical change in microlithic technology contemporaneous with increased droughts and severe wildfires in the region60. In Southern Norway, a similar diffusion event occurred, where pressure blade technology suddenly appeared61. These two studies testifiy to the notion that the environment must have affected the Early Mesolithic populations of Northern Europe.

The falling water levels62 would almost certainly have impacted the fishing and hunting opportunities and may have caused humans to move away in search of viable fishing grounds. This may account for the observed decrease of butchered animal remains discarded in environments conducive for preservation. However, ecological stress caused by increasing temperatures, coupled with potential wildfires in drying mixed coniferous forest, may also have triggered the movement of humans and animals alike.

The presence and continued use of fine-barbed bone points in Northern Germany during the hiatus12,63 suggest that fishing practices were not as disrupted further to the south. While dates from the Swedish west coast; Huseby Klev and Balltorp64,65,66 and a further 10 AMS dates from charred remains from the habitation site of Ålyst on Bornholm (Supplementary Figure 24; Supplementary Dataset 2), also suggest the arrival of new groups there. Subsistence at the Swedish sites focused primarily on marine mammals, fish, and birds67. Unfortunately, the coastlines of the Early Holocene in Denmark became submerged during the Atlantic period due to rising sea levels (see Supplementary Information 7), meaning that Early Mesolithic coastal sites are rarely located and excavated in this region68,69. Consequently, it is not possible to study and compare coastal settlements which could corroborate a shift in economic strategy and subsistence on a regional scale during the Early Mesolithic in this area. However, human exploitation of marine resources along coastal environments in Southern Scandinavia can be inferred from stable isotope data obtained from human remains. Individuals from Køge Sønakke off the coast of Eastern Zealand in Denmark70, Österöd71 and Huseby Klev65,72 from the South-western coast of Sweden are contemporaneous with the identified hiatus. These individuals yielded collagen δ13C and δ15N values consistent with partially or fully marine diet (Supplementary Dataset 2). Indeed the presence of harpoons during the Mesolithic has strongly been tied to the hunting of marine mammals, although the fact that they ended up in the potentially dried up lakes may indicate a more varied use36.

In summary, radiocarbon dates show a hiatus, spanning c. 600 years, separating the two classified types of bone points at c. 10,300 cal BP. The hiatus of dated barbed bone points are also evident in the radiocarbon dates from the habitation sites in Eastern Denmark, in effect dividing the Maglemose into two complexes, displaying distinctly different technologies; fine-barbed bone points and percussion knapping in the Early Maglemose Complex, and large-barbed bone points and pressure flaking in the Late Maglemose Complex. These two complexes represent two radically different material cultures and technological traditions which challenges the notion of a period in relative stasis. We are confident that this surprising gap in the South Scandinavian material will find more parallels elsewhere in Europe.

The cause of this radiocarbon gap is currently unknown, but we hypothesize that climate change caused water-levels to drop and thereby forced the Mesolithic people to relocate and adapt their subsistence strategies, potentially along the now submerged coasts. Another possibility coupled to low water levels could be the increase of wildfires due to an exceedingly dry climate, which may also have forced hunter-gatherers to relocate.

The introduction of pressure blade technology is now inferred to be synchronous with the emergence of large-barbed bone points after the hiatus and may indicate the transmission of influences from Sweden. Species identification by ZooMS and LC–MS/MS indicates that a conscious selection was carried out in regards to the specific species (as well as the preferred skeletal elements, i.e. rib or long bone) for manufacture.

The pressure blade industry which characterises the Late Maglemose Complex is thought to have originated at c. 20,000 cal BP in the area of present-day Siberia/Northern China from where it spread westward to Western Russia73, and subsequently into Northern Fennoscandia39,74,75. Results from archaeological studies on the diffusion of pressure blade technology correspond well with genetic studies from Norway and Sweden. These have shown that the individuals involved in the spread of pressure blade technology, were genetically admixed between Western and Eastern Hunter-Gatherers76, and thus indicate migration routes westwards from Russia into the Scandinavian peninsula while the route into Denmark remains inconclusive75 as it could also have spread through Poland and Germany. Most recently, re-analyses of lithic remains attributed to the oldest sequence at Huseby Klev (deep pit) revealed the presence of pressure blade technologies, which are securely dated to the hiatus and the subsequent decades (10,040–9610 cal BP). Moreover, the same study also reported on aDNA extracted from chewed birch pitch revealing that the individuals were of genetically admixed ancestry77. This novel lithic technology coupled with a changed bone point morphology and the pattern of radiocarbon dates from Western Sweden and Central Scandinavia78 favours the spread of this technology through Scania and into present-day Denmark.

It is hoped that future high-resolution sedimentary studies are conducted to elucidate if and in what ways the local climate affected the people inhabiting this area. Similarly, direct radiocarbon dates of material from classic Maglemosian sites, as well as genetic studies, will hopefully help to illuminate whether the Early and Late Maglemose Complex are also genetically as well as technologically distinct.



The bone points analyzed (n = 127) all derive from Southern Scandinavia (Denmark and Southern Sweden) (see Supplementary Information 1). The bone points were, based on barb morphology and skeletal element used in their manufacture (i.e. long bone versus rib), and to some extent species, divided into two groups.

A subset of the bone point samples was further submitted for AMS radiocarbon (14C) dating (n = 23), protein analysis to determine the species of each artefact (n = 120), and carbon (δ13C) and nitrogen (δ15N) isotopic analyses (n = 19). This has resulted in two distinctive groups extending from the Preboreal into the Atlantic Period, corresponding to the Maglemose and slightly beyond, into the subsequent Kongemose culture.

Radiocarbon dating

We submitted bone powder or bone fragments (mean weight 100 mg) of 23 bone points from Denmark for AMS dating at the Oxford Radiocarbon Accelerator Unit based on their typological grouping. Collagen was extracted from bone powder using ORAU pretreatment codes AF (samples A44111, 436, 487 and 534) and AG (all other samples79. The extracted collagen was combusted, graphitised and dated according to Dee and Ramsey80,81 and Ramsey et al.80,81. Of the 23, 21 were successfully dated and were subsequently merged with 24 published AMS dates8,14,15 as well as seven unpublished dates from Zealand (see Supplementary Dataset 1). Two artefacts (FP1469: 9375 ± 45 14C BP, 9208 ± 55 14C BP, and Brokøb B: 7940 ± 65 14C BP, 7890 ± 65 14C BP) were dated twice. Isotopic analysis of the dated collagen samples was conducted offline using combustion IRMS at the Oxford Radiocarbon Accelerator Unit, as were a further five samples dated at The Tandem Laboratory at Uppsala University for which sufficient collagen remained (Supplementary Dataset 1).

The total of 52 AMS dates was then calibrated to cal. years BP in OxCal v.4.3 using IntCal3 calibration curve16,82. We also compiled 68 published and unpublished radiocarbon dates performed on charcoal and bone associated with habitation sites, including human remains not directly associated with habitation (Supplementary Information 2), and 118 published and unpublished radiocarbon dates from faunal remains (elk n = 33, red deer n = 12, auroch and bison n = 73) spanning the Maglemose to infer presence or absence (Supplementary Information 2). We applied Bayesian phase modelling on the dates from habitations, assuming the coeval age of habitation events using OxCal v.4.382 (see Supplementary Dataset 3 and Supplementary Figure 25).


We performed ZooMS on 120 bone points from Denmark and Scania in Southern Sweden using protocols from83,84 (see Supplementary Information 3). Mass spectrometry was conducted on a Bruker MALDI-TOF–MS/MS instrument in reflector mode to acquire spectra from 800 to 3500 m/z. Taxonomic identification was completed using published markers17. As red deer (Cervus elaphus) and European Elk (Alces alces) cannot be distinguished with published markers85, we compared their collagen (COL1ɑ1 and COL1ɑ2) sequences and identified five single amino acid polymorphisms (SAPs). We then analyzed these SAPs to see if they provided unique tryptic markers for ZooMS analysis (see Supplementary Information 4).

To confirm our candidate ZooMS marker capable of discriminating between red deer and elk, four reference samples (two from each of these species) were digested with Trypsin, Elastase and Chymostrypsin and sequenced using LC–MS/MS (see Supplementary Information 3). In addition, we sequenced the tryptic peptides from four bone points, two from the Preboreal and two from the Boreal (see Supplementary Information 3).