The Neolithic settlement complex at Fannerup

Abstract

The Neolithic settlement complex at Fannerup

About 6,000 years ago, the northern part of Djursland (East Jutland) was an island, after the Littorina Sea had penetrated into the Kolind Sound (fig. 1). Along the old coastlines, which are now raised 3.5-5 m above sea-level (1), are several settlement and shell heaps. They derive in particular from the Ertebølle Culture (EBK), but there are also -as this article will show- settlements belonging to the Funnel Beaker Culture (Tragtbægerkultur TBK).

At the village of Fannerup on the north side of Kolind Sound, 11 shell heaps have been registered (fig. 2). Most of these lay on a ness which du ring the Stone Age was 120-160 m wide and 8-9 m above sea-level. To the north, the ness was linked to its hinterland by means of a narrow tongue of land.

Some of the shell heaps at Fannerup were already known in 1850 (2), and 8 years later A. P. Madsen was the first archaeologist to work there (fig. 3). Also King Frederik VII planned to excavate the Fannerup shell heaps in 1860, but this never materialized (3).

In the years 1888, 1889 and 1891, A. P. Madsen registered 7 shell heaps in Fannerup. He was particularly taken up with one of the shell heaps, viz. Ørum Å, because it contained large amounts of artefacts and animal bones from TBK. The shell heaps known hitherto all belonged to the EBK and since their artefacts seemed more primitive than those found in the dolmens and passage graves, it was among other things believed that the Stone Age could be divided into an earlier and a later part, corresponding to what we today call Mesolithic and Neolithic (4).

Ørum Å deviated further by having a different structure to the EBK shell heaps, which consisted mainly of whitish-grey compact layers of oyster shells. Its culture horizon was very dark and contained only a few shells (5). A. P. Madsen wrote a treatise on Ørum Å, but never published it (6). The investigations which during these years were carried out in different shell heaps were actually probes in preparation for the coming major shellheap project to be carried out by the National Museum, the results of which were published in 1900 (7). In connection with this, the rest of Ørum Å was investigated by C. Neergaard in 1895.

After the big shell-heap publication, Danish research into shell heaps practically ceased, and it is characteristic that no further archaeological investigations were made at Fannerup until the 1950s, when removal of the Fannerup ness in connection with construction work revealed new shell heaps. These three shell heaps, which all belong to TBK, and are called Fannerup I, II and III (FI, FII and FIII) were investigated by P. Kjærum. The presentation of the four Neolithic shell heaps/settlements here (Ørum Å, FI, FII and FIII) is due partly to the fact that the artefact material is chronologically uncorrupted and partly that a large number of bones occurred in these excavations, which are analysed elsewhere in this issue by P. Rowley-Conwy.

FANNERUP I

The FI settlement, which lay on the NE side of the ness, was entirely removed in 1953. P. Kjærum managed to investigate a smaller part of the settlement before this happened (8). FI has an area of 800-1000 m2 (9).

Stratigraphy

The stratigraphy of the 33 m2 excavation area (fig. 4) was documented in 11 profiles, two of which are reproduced here (fig. 5-7).

The up to 1.25 m thick settlement horizon, which lay encapsulated between the sandy subsoil (layer 6) and the topsoil (layer 1), consisted of four different layers: 2-5.

Layer 5 (the bottom layer) was a 2-6 cm thick layer containing charcoal, which extended almost unbroken through the whole excavation area.

Layer 4 (the oyster layer) lay directly above layer 5. The up to 20 cm thick layer consisted almost only of shells, especially of oysters, but also Cardium, Littorina and common mussel were present.

Layer 3 (the shell layer) was suprajacent to layer 4 and was 35-75 cm thick. It consisted mainly of a dark sticky matrix containing large amounts of fragmented shells.

Layer 2 (the dark culture layer) lay above layer 3 and was 30-60 cm thick. The matrix was rather dark and sticky, no doubt due to a large content of decomposed organic material.

Structures

In the subsoil surface were seen a number of pits and stone settings, some of which were fireplaces. A stone-set fireplace at the bottom of layer 4 is seen in fig. 7.

Layers 3-5 contained horizontal 3-6 cm thick clay lenses of varying extent, up to as much as 5 m2. These were no doubt man-made, but it is not known what purpose they served. The intrusions from layer 2 into layer 3 resembling post-holes are erosion phenomena.

Finds

In the presentation of the finds, the main emphasis is laid on the pottery, while artefacts of flint and other stone are mentioned when they have chronological or functional significance. The pottery from layers 4 and 5 is treated as a whole, since the pottery inventory of the two layers is considerably mingled, despite the finds from the two layers being separated during excavation. The artefact material may be mainly dated to Middle Neolithic TBK (TBK MN I-V) with a special concentration in MN II. The datings are supported below.

Pottery. The ratio of vessel sherds to clay discs in the 5413 sherds is shown in table I. For each layer the minimum number of vessels determined from the variation in rim sherds is determined. A survey of the rim and rim-edge ornament (R1-R86) is seen in fig. 8. The vessels are classified into three types: undecorated vessels, storage vessels and finer vessels. The distribution of F I's 208 vessels according to these three groups is shown in table II.

Storage vessels have rim ornamentation like R1-R23 (fig. 8). Rim ornament in the form of chevronlines or -rows is irregularly produced (10), unless they derive from vessels with the characteristic storage-vessel body ornament.

The storage vessels may be divided into two groups: funnel beakers and bucket-shaped vessels. The bucket-shaped vessels are represented by sherds with a coarse, irregular surface, which is very different from that of the other sherds. The rim ornament of these vessels consists either of deep finger impressions (R4, fig. 11g) or a horizontal finger groove (R22, fig. 10j). Only 5 of the 134 storage vessels are bucket-shaped vessels, dated to MN V (11).

The other group of storage vessels, the funnel beakers, may be divided into two classes. In one of these, the vessels have a smooth sigmoid profile in which any body ornament consists mainly of short vertical grooves (fig. 9a-b). In the other class, the vessels have just above the belly turn a little recess which is often decorated with oblique stabbing and occasionally with a row of short, vertical grooves or round stab-marks (fig. 9c-e). The most frequent rim ornament in the funnel beakers is a single row of pits; next comes horizontal chevron-lines and -rows. In addition, 12 other kinds of rim ornament are employed (table III).

About ⅔ of the ornamented side sherds are from storage vessels (table IV). Groove­ornamented body sherds predominate in relation to the oblique-stab ornamented side sherds (table V). In one vessel, the body ornament has been made with a two-ply cord (fig.11h).

The funnel beakers are 10-32.5 cm high (fig. 9a, c) with rim diameters of 12.5-34 cm. The small funnel beakers with a rim diameter of less than 20 cm are not so frequent as the large ones (fig. 12).

The finer vessels are represented by the decorated sherds that do not derive from the storage vessels. Judging from the rim sherds, ⅕ of the vessels are finer vessels (table II).

Layers 4-5 contained 14 finer vessels, 8 of which have horizontal chevron-lines or -rows under the rim (fig. 13a, f, 15a-d). The rim ornament on the last 6 vessels is apparent from fig. 9g and fig. 13b-e, g. Grooves are the dominant technique, but also notch-stamps (fig. 13g, 15d) and two-ply cord (fig. 13e) have been used. Grooves also predominate in the side sherds, but more different techniques than in the rim sherds have been used (fig. 14).

Of the 26 finer vessels from layer 3, 7 have a rim ornament with horizontal rows or lines of chevrons (fig. 16a-b). Horizontal lines or bands are found in 14 vessels (fig. 16c-i, m-r). Two vessels have vertical bands (fig. 16k, 1). The vessel fig. 13b from layers 4-5 is also present in layer 3. Finally one vessel has a horizontal moulding under the rim. Selected side sherds from layer 3 are shown in fig. 17. In both rim and side sherds, grooving is the predominant technique.

From layer 2, only 4 finer vessels are represented in the form of rim sherds. Two vessels have horizontal chevron-lines, and two vessels a horizontal band of horizontal impressions of two-ply cord (fig. 18e-f). Side sherds from layer 2 are shown in fig. 19a-i.

A number of rim sherds of finer vessels have in addition internal rim ornament, which may also occur alone. In certain vessels the decoration is emphasized with an inlaid white mass, see for example fig. 15.

Of the finer vessels, the forms found are among others funnel beakers (fig. 9g), shouldered vessels (fig. 14g, 15a) (12), bowls (fig. 13e-f), and stepped vessels from MN III (fig. 17n) (13). Vessels also occur with a high conical neck, see for example fig. 13a and 14a.

Among the undecorated vessels, identifiable shapes are bowls, a triple-stepped and a two­stepped bowl (fig. 20a-b). One vessel represented by a coarsely tempered rim sherd is from MN V (14).

Clay discs are represented by 180 sherds (table I). Determined on the basis of the rim sherds, there are at least 17 from layers 4-5, 17 from layer 3, and 1 from layer 2 (fig. 21-22). Most are decorated with stab-and-drag, but other techniques, e.g. two-ply cord and Cardium are used (fig. 21 1,n). The ratio of decorated to undercorated discs is 2:3 (15). There is no difference in size and thickness between decorated and undecorated discs (16).

On a large fragment of a clay disc there is a dark circle (fig. 23), a discoloration probably due to a thin coating of carbon compounds (17). How this spot has arisen has still not been determined. The function of clay discs -although they are often said to be baking plates- is far from clear (18).

Artefacts of flint and greenstone. 3 flake axes and 31 fragments of Neolithic axes (table VI) were found (19), not counting some smaller chips with polishing. Fifteen of the axe fragments are fire-brittle. Among the axe fragments is a small piece from a shafthole axe of greenstone (in Danish a stone other than flint, whether igneous or sedimentary).

6 transverse points (fig. 24) were found. Five of these should be dated to TBK (20), while the last must be dated to EBK (fig. 24d). A tanged point (fig. 24h) derives from the Single Grave Culture (21).

Fragments of polished flint chisels, blade sickles (fig. 24a-b), querns and grinds tones were also found.

A registration of the unpolished flint has been carried out only for layer 5, see table VII (22). Locally found flint nodules have been used as raw material. Similar flint nodules were found in the subsoil during the excavation (23). Not all flakes are included in table VII (24). Irregular disc-shaped chips predominate (25). The 26 cores are irregular and small, and the much worked cores can be described as "nodules" (26). One spherical striking­stone of flint (27) and 10 elongated rounded striking-stones of greenstone were found. The raw material for the 50 scrapers are cores and especially disc-shaped flakes (fig. 25). The 9 borers comprise 2 thick borers fashioned from core-pieces and 7 borers made from flakes. One of the borers is a 9.2 cm long shouldered blade borer (28). Among the 15 flakes with edge retouching are 2 A blades and 4 B blades (29). That this last-mentioned artefact group consists of good, regular flakes is further seen in the fact that cortex is preserved in only 2 pieces. This is a very small proportion, cortex being otherwise preserved in 40% of the flint implements. In the group of flakes with transverse retouch are 2 A blades with straight retouch.

Generally, the flint inventory of layer 5 -apart from a solitary flake axe- should be described as Neolithic.

Artefacts of bone and antler. 28 implements of organic material may be referred to a particular layer; only one piece was found in the lime-deficient layer 2. Perforated astragali are the commonest implement form of organic material, 20 fragments (of ox, fig. 26a) and a whole specimen with unfinished perforation (pig, fig. 26b) being found. The function of this implement group is unknown, but astragali may have been used in drilling with a bow drill. 2 awls each with a half trochlea (fig. 26c-d) were also found. One of them is made from tibia of sheep or goat. Both perforated astragali and awls with half trochlea are common at TBK sites (30, 31). A "smoother" of the same type as those assigned to EBK (39) must here be attributed to TBK, since it is probably made of cow radius (fig. 26e). A smoothed, pointed red-deer antler (fig. 26f) has parallels at other TBK sites (33). Finally, a fashioned semicircular bone plate was found (fig. 26g) and a fragment of a red-deer antler with incised grooves (fig. 26h).

FANNERUP II

The Fannerup II settlement (F II) was also exposed when the ness was removed in 1953. A small part of the settlement, which lay in the NW part of the ness (fig. 2.), was investigated by P. Kjærum (34). 6 m2 were excavated at a spot where the thickest settlement layers were found (fig. 27a). From the two northern squares artefacts have not been collected from layer 2.

Stratigraphy

The stratigraphy showed only two prehistoric structures (layers 2 and 3) between the topsoil (layer 1) and the subsoil (layer 4), cf. fig. 27b. The lower cultural horizon, layer 3, consisted of black culture soil with fragmented shells, while the upper culture layer, layer 2, was dark and sticky and devoid of shells. The same sequence was also found in the more elevated part of the settlement (fig. 27c). The two layers may largely be equated with the corresponding layers in F I. In F II, as in F I, there were thin clay lenses and erosion structures resembling post-holes.

Finds

In the following, with one exception, only artefacts from the excavation area itself (fig. 27a) will be presented.

Pottery from layer 3 comprises 237 sherds (table VIII). On the basis of the rim sherds, 17 vessels may be distinguished, 9 of them storage vessels. These vessels are mainly decorated with a series of pits, but finger-tip impressions with nail and double chevron-lines also occur (fig. 28a-c). The commonest body ornament in storage vessels is short vertical grooves, but in two other vessels, vertical Cardium impressions have been used (fig. 28a) and triangular stabbing (fig. 28h). All storage vessels are funnel beakers with a smooth transition between neck and body. The rim diameters lie between 22 and 36 cm.

Only 3 finer vessels were found, two of which are decorated with double incised chevron­lines (fig. 28d, 1). An open bowl is furnished both externally and internally with vertical rows of large arc-stabs (fig. 28e). The last five vessels, among which are two bowls, are undecorated.

Among the 13 fragments of clay discs, a decorated and an undecorated disc may be discerned. The decorated one has a double row of pits around the rim (fig. 28m).

Pottery from layer 2 comprises only 60 sherds. Five of these, including two rim sherds, are decorated. One rim sherd has an incised arc, the other an applied wave moulding (fig. 29a­b). The side sherds are decorated with notch-stamps, grooves and arc-stab lines (fig. 29c-e). There are also small fragments of a rim sherd and two rims of clay discs.

Artefacts of stone and greenstone. A large amount of flint implements and flint waste were found which will not be described. Of axes, only two chips are preserved and a large fragment of a polished thin-butted flint axe. The fragment, which derives from the base of layer 3, may -although but one face and a side are polished- be dated to MN II (35).

Layer 2 contained like layer 3 a regular Neolithic flint inventory, but additionally some typical EBK artefacts and an Iron Age hammer-stone (36).

Finally, an unidentified piece from the higher part of the settlement should be mentioned (fig. 29f). This is possibly a broken blank for a point-butted flint axe with hollow edge. These axes are normally dated to late TBK (37).

Dating

The Neolithic sherd material may be dated to MN II, since the sherds are decorated in MN Il style with a few exceptions: the ornament of sherd fig. 28i is in MN I style, while the sherd fig. 29c must on account of the notch-stamps be dated to MN III-IV.

FANNERUP III

The midden Fannerup III (F III) was found in 1956 when a cellar was being dug for a house. F III's position is shown in fig. 2. During P. Kjærum's investigation of the same year, a 10 m long and 1 m wide trench was excavated.

Stratigraphy

As the profile section fig. 30 shows, several layers could be differentiated. Under a c. ½ m thick layer of topsoil lay the culture layer, which consisted of layers 3, 4 and 5. Layer 3 was a compact layer of shells, while the two others consisted of culture soil mixed with shells. The three layers rested on layer 1, which consisted of sand with a number of shells. This layer contained -like layer 2- no artefacts. Traces of two fireplaces were found.

The pottery comprises 246 sherds (table IX) corresponding to at least 15 vessels and 2 clay discs. One of the vessels is from the Iron Age.

Of the 8 storage vessels, 4 are decorated with a row of pits under the rim (fig. 31a), 3 have incised chevron-lines and 1 is furnished with a row of finger impressions. The storage vessels are otherwise mainly decorated on the body with a row of short vertical grooves, but also trapeze stabbing (fig. 31a) and applied moulding are employed.

The finer vessels are represented by a single rim-sherd only. It is decorated both inside and out with horizontal lines and chevron-lines (fig. 31e). There are also 9 side sherds corresponding to at least 4 different vessels, including one with a large strap-like lug (fig. 31 b) from a shouldered vessel (39) and a bowl with vertical Cardium bands with intervening empty fields (fig. 31c). The sherd fig. 31g was found during the excavation of the cellar prior to the archaeological investigation. It should be assigned to MN III.

The two last vessels have been undecorated, like one of the clay discs. The other clay disc is decorated (fig. 31f).

Flint artefacts. The flint inventory is a mixture of implements and flakes from EBK and TBK. The Neolithic axe material contains only a couple of polished flakes and a fragment of an unfinished thin-bladed flint axe, whereas no fewer than 16 flake axes and one core axe of the EBK, which is also represented by other types of artefacts, were found (40). Flake axes are not part of the artefact inventory in TBK MN (41), but are intrusive from previous EBK occupation. It is also characteristic that many flake axes are found in the low-lying parts of MN settlements (F III, Ørum Å 1888-89), while in the more elevated parts of the settlements there are no or only very few flake axes (F I, F II, Ørum Å 1895). The designation "flake axe" is perhaps unfortunate, since wear analysis suggests that they have been used mainly in scraping hide (42).

Other finds. A bone awl and a presser of antler and 4 small pieces of clay daub were found in layer 4.

Dating

The compact shell layer, layer 3, is Neolithic, among other things because a crushed Neolithic clay pot was found there. Layer 4 is Neolithic, since it overlies layer 3. Layer 5 is dated -like the other layers- also to TBK MN, on account of the presence of Neolithic vessels. The pottery must be dated mainly to MN II, although the sherd decorated with notch-stamps, fig. 31d, cannot be dated more precisely than to MN II-IV.

ØRUM Å

The Ørum Å midden, which lay on the west side of the ness (fig. 2), was, as mentioned in the introduction, excavated by A. P. Madsen in 1888, 1889 og 1891, and by C. Neergaard in 1895 (43). The areas they excavated are shown in fig. 32. As Ørum Å is mentioned at several places in the literature (44), the present account will be confined to orientation, merely providing supplementary information on certain aspects.

Stratigraphy

The 10-95 cm thick culture layer has been described as consisting mainly of a dark mixture of topsoil, ash and charcoal dust, with a scattering of fragmented shells that never formed continuous layers. At several places there were the remains of fireplaces (45, 47). A photograph of a profile section shows a rather heterogeneous formation with several layers which have not been plotted, however (46). The lowest-lying part of the Ørum Å settlement rested on a disturbed heap of shells containing a number of EBK artefacts (1888-91 excavation).

Finds

The find material from Ørum Å is a copious and rich settlement material from TBK. In the elevated part of the settlement (1895 excavation), the only certain EBK artefact was a flake axe. Distribution analyses of flint waste and sherds show that the artefacts are evenly distributed in the culture layer. The sherds from the same vessel are found at entirely different levels and spots (fig. 32).

Pottery. The treatment below of the pottery comprises only the material from the excavations i 1888, 1889 and 1895. A few sherds from these excavations could not be located, however (48). The minimum number of vessels and clay discs is apparent from table X, while the rim ornaments are shown in fig. 8. The larger part of the sherds are decorated in MN II style. Two sherds are decorated in MN I style (fig. 8: R40, R42) and one vessel in early Ferslev style (fig. 34). The artefact material treated here may be dated to MN II (48, 51).

Shaft-hole axes of greenstone. An entire shaft-hole axe, a butt fragment and an edge fragment were found at Ørum Å (fig. 3). K. Ebbesen has constructed a typology for shaft-hole axes from TBK MN. Here two of the Ørum Å axes are assigned to type A2, which is dated to MN I-II (49, 50). However, my measurements of axes' cross-sectional index show that the axes must belong to Ebbesen's type AI, which he dates to MN Ia. The axes must, like the rest of the artefact material from Ørum Å, be dated to MN II, so Ebbesen's oldest type (AI) is not sufficiently narrowly defined.

DATING

TBK MN is on the basis of the settlement pottery divided into five periods (MN I-V), where MN I is further sub-divided into MN la and Ib (52). In addition, the Fuchsberg phase has been defined at the transition between EN and MN (53). A period may manifest several ceramic styles, so in some cases MN Ib style may be dated to MN II. The different styles found in the sherd material from the four Neolithic Fannerup settlements are shown in table XI together with the dated flint artefacts.

MN Ib style. Examples of this style are shown in fig. 18f and 28i, cf. also fig. 8: R40-42 (54).

MN II style. Nearly all the sherds from the Neolithic Fannerup settlements are decorated in this style. Thus all the funnel beakers among the storage vessels have features characteristic of MN II style. This namely applies to the body ornament's rows of short vertical grooves (fig. 9a-b) (56). The funnel beakers belong to Becker's E group, which is dated to MN II­III (55). The employment of different stab techniques on the rim and body of the same vessel (fig. 9d, 10f) is also known only from MN II vessels (57).

In the finer vessels there are many elements characteristic of MN II style: I) the narrow shoulder of the shouldered vessels (fig. 15a) (58); 2) the funnel-shaped neck and lack of handles in the shouldered vessels is a local feature in the MN II of north Jutland (59); 3) the bi-conical vessels (e.g. fig. 14a) have parallels in the Walternienburg Culture (60); 4) internal chevron-lines on bowls (fig. 16b) (61); 5) vertical line groups under the rim ornament (e.g. fig. 14a) (62); vertical ornament groups in various executions from neck ornament to the neck-body transition are known in MN II-III style, so those that have not been made by notch-stamps may be dated to MN II (e.g. fig. 15b, c, e) (63, 64); 7) body ornament may consist of alternating fields in which the decoration is differently oriented (fig. 14b, c) (65); 8) single or paired standing chevrons at the neck-body transition (fig. 9g, 13a) (66); 9) in rim ornamentation, chevrons are mainly incised and arranged in rows. Stab and drag is much employed. Finally, side stamps with incised chevrons or lines is typical of MN II style (e.g. fig. 15b) (67).

All the decorated clay discs may be dated to MN II. This dating is in accordance with the small size of any holes in the discs (68).

MN III style. The MN III style of north Jutland is called the Ferslev style and is characterized by all the ornament being almost exclusively made by notch-stamps. A distinction is made between early and late Ferslev style corresponding to MN II and MN III, respectively (69). The early Ferslev style occurs in vessels that are morphologically older than MN III, and in execution of the ornament other techniques besides notch-stamps have been used.

Three sherds (fig. 13g, 17n, 31g) are dated to MN III, since they belong to a group of vessels whose dating basis has been precisely defined (70).

One vessel (fig. 34) from Ørum Å has previously been dated to MN III (71), but since the style is early Ferslev, the vessel should be dated to MN II (72). It also lay scattered in the MN II dated culture layer (fig. 32).

MN IV style. Incised triangles filled out with small stab-marks or consisting entirely of these (fig. 19e, f) (74).

MN V style. Six sherds (i.a. fig. 10j, 11g, 18g) must be dated to M V, since they derive from the large bucket-shaped vessels typical of the period. The vessels can be decorated either under the rim with a horizontal finger groove or with a horizontal row of deep finger impressions (75).

MN II-IV. A number of sherds (i.a. fig. 16r, 17m, o, p, 29c, 31d) cannot be dated more precisely than to MN II-IV. The sherd fig. 19c is probably from MN III, although the pattern also occurs in MN IV (73).

Dating of F I-III and Ørum Å

The dating of the individual sites and layers will be apparent from table XI. The main occupation in all sites lies in MN II, and after this there are signs of briefer occupation in MN III-V, Single Grave Culture and Iron Age. The 4 sherds in MN I style should probably be dated to MN II (76).

There is a clear difference in ornamentation in the finer vessels in the two MN II dated layers F I, layer 4-5 and layer 3. Notch-stamps in rim ornament are more frequently used in the younger layer 3, corresponding to the style development towards late Ferslev style (fig. 8) (77). The late Ferslev style is also characterized by a more extensive use of horizontal bands or parallel lines under the rim. This is reflected in the material in that they are found in 12 vessels from layer 3 and none from layer 4 (78).

A multivariate correspondence analysis of the sherd material from the layers in the four Fannerup settlements showed no chronological differences, from which it cannot be deduced, however, that they are contemporaneous (79).

If the pottery from Ørum Å is regarded as a single chronological entity, it must be considered contemporaneous with F I layer 3 and younger than F I layer 4-5.

C-14 dates. K-4049. F I layer 5: 2440±70 BC.

K-4050. F I layer 4: 2470±95 BC.

K-4051. F II layer 3: 2400±65 BC.

All dates are uncalibrated (80) and are in accordance with other C-14 datings of MN II material (81).

THE CHARACTER OF THE SETTLEMENT COMPLEX

Since less than 1% of the Fannerup ness has been investigated, it is difficult to evaluate the character of the settlement. Older reports from the time before the destruction of the ness suggest that large parts of it have been covered by prehistoric culture layers (82).

F I-III and Ørum Å must be described as settlements with thick culture layers deposited on the spot. The clear stratification of, for example, F I testifies to several successive occupations. The most characteristic layer formation is dark culture soil with fragmented shells, which must have been formed by the accumulation and breaking down of large amounts of organic material. Massive layers of shells of oysters and other molluscs occur only in F I and F III, and then in very thin layers, which are very different from the thick layers in EBK's middens.

The artefact inventory is a regular settlement material. The pottery, too, is clearly utility, 60 % of the vessels being storage vessels, for instance, and less than a quarter finer vessels (83).

The location of the four sites on a ness surrounded by salt water must have been ideal for hunting, fishing and gathering, but this does not seem -despite the recovery of two presumptive fish-hooks from Ørum Å (84)- to have occurred to any great extent. P. Rowley-Conwy's analysis of the bone material from F I-III (this KUML) shows that seal, red deer and other game comprise less than ⅕ of the slaughtered animals. The Neolithic Fannerup sites do not belong -despite their location- to TBK's hunting stations (85).

The position of the settlements on the ness apparently laid restraints on an agrarian economy. The catchment area that could be reached without crossing water covers within a 4 km radius only a quarter of the total area (fig. 35).

However, the location may have been favourable for cattle-raising, which has comprised about half the animal husbandry. The meadows around the ness have provided good grazing, especially after the regression following the Sub-Boreal transgression (86). The great Stone Age landnam that occurred at the transition between Early Neolithic and Middle Neolithic (87) opened up the forests for cattle foraging and fields. From this landnam on, cattle-raising seems to have played a more important role in the economy (88).

A Neolithic peasant society practising extensive cattle husbandry has thus had good conditions on the Fannerup ness. The main utilization has apparently -judging from seasonal indicators in the bone material- occurred in the winter (89).

TBK has also left its marks outside the ness, where 26 megalithic graves have been recorded within 4 kms (fig. 35). In addition there are the finds of Neolithic artefacts at the top of the three shell heaps from EBK (90).

When interpreting the nature of the settlement, one must be aware that the Fannerup ness may have been a Neolithic causewayed camp (91). Both the topographical conditions (the marked ness partially surrounded by water) and the size suggest this. The settlement in MN II would in this case be later than any camp and be a kind of continuation.

This theory is not supported by the excavation results, but the local school-teacher observed during the construction works of 1949 something which may have been a ditch in a causewayed camp: "Further up the hill, the workmen have come to a large burnt surface about 1 m down. It is 4 m wide and 10 m long or as much as has to date been dug away. I have this afternoon been down to the site in order to dig through the patch, but it was apparently bottomless. There are quite large amounts of undisturbed shell heaps on both sides of the excavated area ... " (92).

ACKNOWLEDGEMENTS

Poul Kjærum, Forhistorisk Museum, is warmly thanked for making available the material from Fannerup I-III and for inspiring discussion on the Fannerup complex.

I am also grateful to Dronning Margrethe II's Arkæologiske Fond, which provided financial support for the preparation of this article.

Palle Eriksen