R . De C eunynck and H. Thoen

T H E IRO N A G E SETTLEM EN T AT D E P A N N E -W ESTHOEK. ECOLO GICAL A N D GEOLOGICAL CONTEXT

R é s u m é . D ans le cadre d ’une étude géologique des dunes de La P anne-W eslhoek p a r R. D e C eunynck une an alyse palynologique d ’une couche húm ense a pu ê tre réalisée. Une datation au carbon e-14 place ce lle -c i de 26 6 0 ± 100 B.P. (à la hase) et 1965 ± 1 / 0 B.P. (au som m et). C ette datation concorde avec le peu de m atérie l archéologique d e l ’A ge du Fer qui f u t récu pérée dan s ce contexte. H. Thoen 2 en treprit l'é tu de de ce m atériel.

Une occasion unique se présen ta it d ’approfondir a insi notre connaissance d e l ’écologie de l ’h abita t préh istoriqu e à L a Panne, site connu dan s les m ilieux archéologiques, m ais à propos duque! on ne sa it en fa it que peu de choses.

I n t r o d u c t i o n (H.T . ) 3

The archaeological site in the dunes o f De Panne has been discovered in 1886 by G. D onny, a Public W orks engineer (D onny, 1886). The initial finds w ere m ainly R om an, bu t soon older objects cam e to light. The discovery o f coins - Celtic, R om an and even sceatias - gave the site a w ide publicity. A t the tu rn of the century the dunes w ere system atically com bed by am ateurs and gradually attracted the attention o f professional circles (de Loë. 1901-2).

The first general survey o f the finds from De Panne w as published in 1902 by Baron A. de Loë, curator o f the M usées R oyaux du Cinquantenaire a t Brussels (de Loë, 1901-2). From this publication it transpires th a t the finds w ere alw ays m ade in one and the sam e area, situated c. 2,800 m. N .W . o f the church o f Adinkerke, som e 200 m. from the Belgian-French border and approxim ately 300 m. from the high tide m ark. This site w as later to be indicated as “Panne I” by E. R ahir. In the years 1902-4 the dunes w ere further explored by J. M aertens de N oordhout. M. de M aere d ’A ertycke and G. Cum ont. This led to the discovery o f a new find-place — “Panne II” - situated in the Rietpanne, better know n as “cam p rom ain" or “Rom eins kam p” . This site lay to the S. o f Panne I, at approxim ately 500 m. from the coast (M aertens de N oordhout, 1903, 157 ; de M aere d ’A ertrycke, 1905, 36). A system atic prospection in 1905-6 by the M usées royaux du C inquantenaire , directed by A. de Loë, resulted in the discovery o f a th ird find-place, som e 350 m. to the S. o f Panne II. This site - “Panne III" - stretches across the border and is

1 Bursaal I.W.O.N.L., Laboratorium voor Fysische Aardrijkskunde, R.U.G.2 Werkleider, Seminarie voor Archeologie. R.U.G.3 For the geomorphological terminology see Depuydt 1966 and 1972. For the chronology o f the La Tène period w e use the amended Déchelette system (cfr. De Laet, 1979, 550-551) :

La Tène I : 500/450-200 B.C.La Tène II • 200-100 B.C.La Tène III : 100 B.C.-early Roman period.

-Pann

Fig.3

WESTHOEK

Bray-Dunes!Ad in kerkeW-Æ. P o ld e rs

G hy vel d e.

GIUM

2ü00m1000

B C5 oooooo

Fig. 1 . - General outline map with location of the map o f fig. 3 and the section-line of fig. 6, 7 and 8 (after Herbauts, 1971 ; Lebbe and De Ceunynck, 1980).

1. Younger Dunes.2. Older Dunes,3. Polders (marine deposits o f Dunkerque-2 age).4. De Moeren.5. Northern border of the older dune sediments.6. Southern border o f the older dune sediments.7. Section-line o f fig. 6, 7 and 8.

THE IR ON A CE SETTLEM ENT A T DE PA NNE- WESTHOEK 23

partly situated on the territory of Bray-Dunes (Dép. du N ord , France). In 1907, G. C um ont published the coins (Cumont, 1907). This study m arked the end o f a first period o f active archaeological research.

After W orld W ar I De Panne attracted once m ore the attention under the im pulse o f som e am ateurs, am ongst w hom K. Loppens deserves m ost o f the credit (Loppens, 1928 ; 1932, 95-98). The interest o f the M usées Royaux w as again aroused in 1926 by the discovery o f an u rn burial in Panne I. This resulted in quite an extensive excavation by E, R ahir in 1927 on this site, soon to be extended to Panne II and III and lasting well into 1929 (Rahir, 1927, 1928 and 1930). W ith this large campaign, the official research in De Panne cam e to an end. Although quite a few am ateurs and societies rem ained active, a co-ordinated scientific research w as never achieved. Only part o f the finds from the excavations by A. de Loë and E. Rahir has been published. The m ajority o f the objects have been put in the reserves o f the M usées Royaux in Brussels and have never seen daylight again.

The dunes o f De Panne have been classified as a nature reserve (“De W esthoek") by law on 29.8.1957 and this pu t an end to all archaeological activity there. In 1961, P. Favorel has published a very thorough status quaestionis of the archaeological research in De Panne (Favorel, 1961 -2).

•** *

Our present know ledge o f the De Panne find-place is thus based on the activities o f A. de Loë in 1905-6 and E. R ahir in 1927-9. The poor state o f the soil science, already repeatedly criticised contem porary (e.g. Loppens, 1932, 97-8), resulted in w rong interpretation and registration o f the archaeological layers and traces, w hich w ere all determ ined as "'foyers d ’habitat" and “foyers industriels". The data concerning the house types and even the grave types based on the observations of de Loë and R ahir (e.g. M arien, 1952, 410 ; 1980, 32-34 ; De Laet, 1974, 466 ; 1979, 573-574) should consequently be treated w ith the utm ost care.

The tw o m ain problem s w hich have been the subject o f discussion during the excavations as well as today concern the dating and the character o f the archaeological site.

Chronologically, tw o m ain periods of settlem ent can be ascertained, namely during the Iron Age and in the Rom an period. The Gallo-Rom an rem ains can be dated from the Flavian period onw ards till in the th ird quarter o f the th ird century, i.e. approxim ately betw een 70 and 270 A.D. (Thoen, 1978, 194-202).

On the Iron Age phase o f the settlem ent the archaeologists disagree. Taking the motifs o f pottery decorations as a criterion, A. de Loë and E. Rahir m ade a differentiation betw een “types de prem ier âge du fe r " (Hallstati) and "types de deuxièm e âge du fe r " (La Tène) (Rahir, 1927, 33-50 and Fig. 8-15 ; 1928, 252-255 and Fig. 117-120 ; 1930, 23-30 and Fig. 8-13). The slightly angular form s, m ainly finger-indented (“decoration au doigt") and pinched (“decoration à la pincée") w ere dated as Hallstatt, the sharply angular pots and bowls, usually com bed or fu rrow ed (so called cardium -decoration) (“décoration au peigne et linéaire"), as La Tène.

24 R. D E CEUNYNCK AND H. THOEN

M. E. M ariën excludes an E arly Iron Age (Hallstatt) phase. He dates all pre- Rom an earthenw are in the Late Iron Age (La Tène). thereby m aking a distinction between E arly La Tène w are (sharply angular forms), due to their relation w ith the carinated M arne w are (M ariën, 1952. 377-380 and Fig. 352-353). and Late La Tène ware, com prising the m ore rounded form s, w hich w ould have superseded the angular form s (M ariën, 1952, 410-412 and Fig. 369-370). As regards the decoration, M ariën suggests th a t the earliest types w ould have been sparingly decorated, w hereas the later ones w ou ld have been abundantly finger-indented, pinched, com bed, etc. : especially the latter decoration w ould have been typical o f the La Tène III period. Like A. de Loë and E. R ahir, M ariën also stresses the continuity betw een the Iron Age settlem ent and the R om an period (M ariën. 1952, 4 1 0 ; 19 7 1 , 220 and 223-224). In his 1971 study M ariën clearly looks for a relation betw een the different archaeological groups in Belgium and the location o f the different tribes k now n from literary sources (M ariën, 1971).

H ow ever, du ring o u r research in to the G allo-R om an settlem ent in the Belgian coastal plain, w e ascertained th a t com bing w as precisely the characteristic m ode o f decoration o f the local G allo-Rom an pottery, and that the com bed w are dated by M ariën as Late La T ène w ere in fact G allo-Rom an pottery products from the 2nd.- 3rd. century A .D . (Thoen, 1978, 97 and 179-188). Subscribing to this view . S. J. De Laet rejects the continuity betw een the settlem ent in the Late La Tène period and the R om an period, a theory th a t form erly had been generally accepted (De Laet, 19 74 , 465 ; 1979, 573). H ereby he relies on the fact tha t the Gallo-Rom an inhabitation in D e Panne starts only around 70 A.D . (research Thoen, 1978, 194- 202) and on the presence o f a sterile layer o f drift-sand betw een the Iron Age and th e Rom an stra ta (research de Loë, 1908, 37 and Fig. 3). In contrast to M ariën. De Laet stresses an E arly La Tène settlem ent a t De Panne, belonging to La Tène 1 and II (De Laet, o.e.).

The character o f the site has been interpreted from the beginning in the light of the briquetage finds. These com prise the earthenw are cylinders or “nails , tripods, perforated plaques, etc. T hey w ere determ ined by A. de Loo and E. R ahir as technical im plem ents o f a potter, hence the in terpretation o f their find-places as 2foyers industriels" or “foyers de fabrication (de poteries)'' in contrast to the /ovens cl'habitat“ or h u t com plexes. E. R ahir even w ent as far as publishing a substantial study, entitled “L 'Age du Fer à L a Panne. Une fa b riq u e de P oteries' (Rahir, 1927), ^ h e re in he treated all aspects o f the production and w hich he lavishly illustrated ^ i t h reconstruction draw ings, chem ical analyses, etc. This nice theory collapsed hke a house o f cards w hen K. Loppens indicated the discrepancy betw een the finished products (Iron A ge earthenw are) and the raw m aterial (post-Rom an D unkerque-2 clay) (Loppens, 1932, 96) and w hen it becam e know n that briquetage w as not related to potting bu t to salt-m aking. H ow ever, it w as not before 19 6 1 th a t De P anne briquetage w as determ ined as such by J. N enquin an K - R iehm (N enquin , 1961, 9 3 -9 5 ; R iehm , 1961, 189). The form er au thor suggested th a t salt-m aking took place both during th e Iron Age and the Rom an P eriod , since hriquetage had been found in layers from those periods. F rom the

THE IRON A G E SETTLEM ENT A T DE PANNE- W ESTHOEK 25

num erous publications and from the excavation reports by A. de Loë and E. Rahir it appears how ever that briquetage has been found predom inantly in Iron Age contexts. The few tim es it w as present in R om an pits, the briquetage has always been found in association w ith Iron Age pottery sherds. W e th ink that these finds are rather to be explained as having been disturbed by Rom an activities and that no p roof has yet been proffered for the use o f briquetage - and therefore o f salt- m aking - in the R om an period at D e Panne (Thoen, 1978, 50).

As to the physical-geographical aspect o f the settlem ent, we have also put forw ard the view that the Iron Age habitation developed on a beach flat protected by dunes (Thoen, 1978, 57-59) (Fig. 7). Thereby w e have rejected the theory on the evolution o f the dunes advocated by the geom orphologist F, D epuydt (1966 and 1972) (Fig. 6). As determ ining factors for the im plantation and the developm ent of the Iron A ge settlem ent at De Panne w e have cited the safe location behind the old range o f dunes, the proxim ity o f the sea (im portant for salt- m aking and fishing) and the presence o f an old river-m outh to the E. o f the site.

W ith respect to the dating problem we have also pointed out a possible relation betw een the developm ent o f the settlem ent and the active phase o f the D unkerque- 1 transgression. Since De Panne w as one o f the few sites in the old dune area w hich w ere n o t flooded, its advantageous geographical position m ust have been attractive to prospective settlers, especially during the active phase o f D -1. The latter w e have dated from c. the 5th. to the 2nd. cen tury B.C., relying on inform ation from the N etherlands, and archaeologically this coincides w ith the La Tène I and II periods. For the sam e reasons the salt-m aking com m unity, w hich w as closely linked to the proxim ity o f the sea, w ould have been on the decline or even been com pletely absent at the end o f this transgression. Hence the few rem ains or even the com plete absence of La Tène III rem ains (Thoen, 1978, 58- 59). Follow ing this line of thinking, the dating o f the Iron Age settlem ent proposed by De Laet w ould be correct, in contrast to the one suggested by M ariën.

It is against the background of these latest opinions on the character and date o f the De Panne Iron Age settlem ent that a study has been m ade o f a hum ic layer, discovered in 1978 by R. De Ceunynck during a geological survey o f the dunes.

T h e S i t e (R.D.C.)

The hum ic layer w hich contains archaeological rem ains w as found som e 20 cm beneath the surface o f a little dune depression deprived o f vegetation (Y on Fig. 3). T he elevation o f this depression is approxim ately + 5 m (Zero H Oostende).

W hen one w ants to com pare the position o f the new find-spot w ith the excavations o f A. de Loë and E. R ahir {cfr. supra), it m ust be recognized that the exact location o f those excavations is unknow n , because only a sim ple draw ing has been m ade by them (Fig. 2). This draw ing indicates only the m ost apparent geom orphological caracteristies of the area. T he distances on the m ap are severely distorted. N evertheless one can conclude that it is highly probable that the find- spot Panne II w as then partially exposed to w ind erosion in the deflation area o f

R. DE CEUNYNCK AND H. THOEN

LAISSE DE HAUTE MER

I M P

f f i - . . y y E g o a .

<-> Panne IQ

Panne I

& V

¿d PanneH 200m

SONDAGE FOYER RECONNU

□ G ISEMENT FOUILLE

F ig . 2. - Scetch-map of the excavations by A. de Loë and E . Rahir (after Rahir, 1930).

THE IR ON A G E SETTLEM ENT A T DE PA NNE- WESTHOEK

+ 17 5 ^ = ' / < -1 1 *

s' . i i i k ̂ 11 > 11 A\ ̂ J* « » /V>

- > v " » i f n̂ o N \/^ <\V/> <A*v \ '> ///

^ ^ N v i P p n n p * v v / ;

r* C C '7‘," ‘ "1 ’‘.V5¿ J - Z' ‘ï ' ƒ

V A /» / I m N '// \ \S A v ^

nVXA- \V\ O, j

** v>sA/ \^/v, Pannell /J r 'w . / " M / ,

V onW ,,,>ví? - _^ - 'y

N

' / m u ' 0, , " , /i(, '""M n> **

V/</

V =’*12,5 ' 1,l * / «v

, p 1%'f aV ;/ •NvV/-r’,,,,,"sV/i

* ov' y P 7 " " r'> " " " " "/V,Mo ',M vA/w vA/VAN 1 1 u " * A /¿-̂ j* i ‘11 \W /S

'V ,” ,,Mi;- xNN^-.n’* o‘,v „,o

ie \ f r t N\V/v V /> ' / //_‘ i" '1 \NN/>

'SS'V'/ ' ' 1 /!f/ ' " o '1 Vi#Ar ■/,„ o '

A r \ v y"'*ST< -, u \ l / t , \ s \ / / , v>NA r o ' ï ' “ ' .Jr ' 0 >’’" '

O „i"»""' » 1111.it" ' OWN/,

V o* '* 0 P a P n e ® ■-< <\X <«> V\V/z ,W - ^ » l i i

oL

5 0 0 m

Fig. 3. - Location o f the excavations by A. de Loë and E. Rahir (Panne I, II, III) and the present find-spot (Y) on a modern map.

28 R . D E CEUN YNCK AND H. THOEN

the m ost w esterly parabolic dune. In the m ean tim e, the vegetation expanded in eastern direction, pursueing the shift in the sam e direction o f the parabolic crest over 150 m eters (Christiaens, 1976). Thus the site becam e covered with a dense dune scrub o f Hippophae and Salix . The proclam ation o f the dune area as a natural reserve in 1957 also facilitated this natural process by prohibiting further archaeological excavations. Som e rem ains o f the excavations at Panne I are still visible at this time. Here the vegetation consists o f low herbs (mainly grasses) Preventing the total disappearance o f indications o f archaeological activity.

As a resu lt o f the previous findings the approxim ate location of Panne I, II and Hi could be defined on a m odern m ap w ith the aid o f aerial photographs (Fig. 3). The position o f Panne III is still uncertain . W hen one considers the distribution of the fmd-spots it seems likely tha t these spots are part o f one large site,extending Underneath the local topography.

Sect ion 1 S e c t io n 2 S e c t i o n 3

30-

eo-

2 0 -

30-

5 0 -

60-

30-

50-

H g . 4 .

Polien-ar\alysis (section 3). 1.2 .

3.4.5.6 .

10 50cm

De Panne-Westhoek. Profiles, showing sampling-spots for C-14 (section 1 and 2) and

Tellow-grey dune sand.Humic horizon, dark grey.Tellow _grey dune sand.j^ark grey to black humus-rich sand ; small sherds.

V,ark grey-brown humic sand ; numerous larger fragments of earthenware. e llow-brown dune sand.

D e s c r i p t i o n o f t h e P r o f i l e s (R.D.C.)

T h eth ' tlery fragm ents w ere found in a hum ic stratum w hich w as about 40 cm ]yu lo ca tio n : see fig. 3). The upper part o f it (layer 4) contains m uch m ore o b s ^ ^ an l° w er one (layer 5). The greatest concentration o f ceram ics w as

e rVed in the latter, it w as m arkedly less in layer 4, w here the fragm ents w ere

THE IR ON A G E SETTLEM ENT A T DE PA NNE- WESTHOEK 29

also m uch sm aller and fragile. The hum ic layer is covered w ith an overburden o f 20 cm dune sand (layer 1 to 3) w hich show s a few hum ic horizons at the level - 0 ,1 0 m (layer 2). D une sand also form s the substratum (layer 6).

Three sections w ere taken, each only a few m eters apart. Section 1 w as sampled for C-14 analysis at the level - 0 .2 5 m, section 2 for C-14 analysis at the level - 0.65 m. Section 3 w as continuously sampled for pollen analysis w ith the aid of small metal boxes.

The hum ic layer appears to be very sandy and rather hom ogenuous. A lthough w e think that at the onset the layer originates from a soil-forming process, further grow th o f the layer m ust have been influenced by m an. This is clearly indicated by the presence o f the potsherds. In view o f the fact that m ost sherds w ere found in the low er part o f the layer, hum an influence w as perhaps greater in the first period (represented by layer 5) than in the second period (layer 4).

PO LLEN A N A LY TIC A L IN V ESTIG A TIO N (R.D.C.)

introduction

All the levels studied pollen-analytically have com e from undisturbed continuous samples. Preparation in the laboratory was executed according to the abbreviated acetolysis m ethod, preceded by boiling in diluted fluoric acid.

Since in a coastal dune area open vegetation plays an im portan t role, a basic pollensum w as chosen w hich gives an indication o f the relationship between fo re s t shrub-vegetation and open vegetation. In consequence the basic pollensum includes trees, shrubs and terrestrial herbs. Spores and Aquatic pollen are not included (Fig. 5).

The m ain diagram A show s the relationship betw een trees, shrubs and terrestrial herbs. D iagram B gives som e specific curves o f individual pollen-types included in the basic pollensum . D iagram C show s the curve o f A quatic pollen. The Spores are not represented in the diagram because o f their very limited presence. M ainly spores o f Dryopteris-, Sphagnum - and Equisetum -type w ere found.

W hen w e look at the diagram there is no apparent indication o f disturbance o f the pollen containing layer. This agrees well w ith the fact that no sign o f digging activity w as found in the cuttings. Instead the m ain diagram A show s a typical transition from open herbaceous vegetation to dune scrub (cfr. infra) as w itnessed in o ther diagram s o f the coastal area.

Zonation

Zonation has been executed according to the “polten assemblage zone concept” (M oore and W ebb, 1978). The use o f this system m eans tha t one zones each diagram upon its ow n features w ithout reference to any other pollen diagram and w ithou t prior attem pts at regional synthesis and correlation. Since som e pollen-

Litlio/ogy

1. Dune sand.2. Very hurrlic sand.3. Slightly humic sand.

Shading of the curves.

4. Trees.5. Shrubs.6. Herbs.7. Gramineae.8. Aquatic pollen.

d e p t h b e l o w s u r f a c e i n ci

l i t h o l o g y

C U d a t i n g

m

m m

s : rrr

Coryl u s

P l a n t a g o s p e c i e s * c o r o n o p u s

P o l l e n s u m ( d i a g r a m A a n d B)

Pollensum.

Diagram A : all pollen except Aquatic pollen and spores.Diagram B : as for diagram A : some specific curves.Diagram C : pollensum diagram A and Aquatic pollen equals one hundred percent.

Fig. 5. - Pollen diagram o f layers 4 and 5 (section 3).

THE i RON A GE SETTLEMENT AT DE PANNE- WESTHOEK 31

types are locally derived, it is possible that a pollen assemblage zone will then simply refer to the state of the vegetation in the immediate vicinity. This is of value w hen one is concerned with elucidating the local succession in vegetation (Moore and W ebb, o.c.) as is the case here.

Zone characteristics :

Zone W a : predom inantly herbs ; the proportion of trees and shrubs is limited to a small am ount (Quercus 4 to 11% , Juniperus 1 to 3 % ).

— Subzone W a1 : dom inated by Gramineae, yet limited im portance o f theCyperaceae and the Aquatic pollen.

— Subzone W a2 : dom inated by Asteraceae liguliflorae (up to 65 96 ), low er valuesof Cyperaceae and above all Aquatic pollen.

Zone W b : at first a sharp incline o f Juniperus then followed by an increase o f Quercus ; higher values of Cyperaceae and Aquatic pollen.

— Subzone W b1 ; increase of Juniperus while the proportion o f Quercus roughlystays the same.

— Subzone W b2 : m axim um of Quercus (2296) while the proportion of Juniperusstays the same.

Zone W c : drastic decline o f trees and shrubs, sharp increase of herbs mainly due to higher values of Asteraceae liguliflorae ; m arked decline of Cyperaceae and especially Aquatic pollen.

Ecology

The diagram is dominated by the herbs, mainly due to the im portance o f Gramineae and Asteraceae liguliflorae. Other herbs that were found are Chenopodiaceae, Asteraceae tubuliflorae, Plantaginaceae and Cruciferae. also - yet sporadically - Caryophyllaceae, Ranunculaceae, Rosaceae, Ericaceae and Papilionaceae (up to 0.5 96) (not represented in the diagram). Higher values o f Cyperaceae probably point to w etter conditions, representing w et till semi-aquatic dune slacks overgrow n w ith sedge. In this context it is w orth noting that the similarity with the curve o f the Aquatic pollen is quite conspicuous. On the other side, the Asteraceae liguliflorae reach their m axim a w hen the values of the Cyperaceae and Aquatic pollen are minimal. This points to drier circumstances and probably even to higher aeolian activity.

Tree-pollen is characterised by high values o f Quercus. Low and continuous values of A Inns, Betula and Corylus were also found. Fagus is rather scarce and has not been found continuously. For this reason a pollen-analytical dating is not possible. This was also the case in m any diagrams in the Dutch coastal dune area (Boerboom and Zagwijn, 1966 ; Jelgersm a et al., 1970). The im portance of the Quercus-curve points to w oody patches of Quercus in the vicinity, possibly w ith a sim ilar composition as the so called Dune Oakwood (“D uin Eikenbos") of The Netherlands (Kalkhoven et al., 1976 ; Doing, 1974). The developm ent o f this vegetation depends on the climate and in the coastal region above all on the wind

32 R . DE CEUNYNCK AND H. THOEN

intensity. As this intensity declines landw ards such a w oody vegetation develops m ostly at some distance from the coast (Kalkhoven et a l., 1976).

During a certain period a dune scrub, dom inated by Juniperus , was present. Sporadically Sa lix was found. Hippophae rhamnoides, the predom inant species of contem poraneous dune scrub in Belgium, was not found.

The possible presence of w oody patches o f Quercus, the presence o í Juniperus scrub and the absence o f Hippophae points tow ards a less calcareous character o f the dunes at that time. It is w orth noting that the hum ic layer appeared to be decalcified and that the dune sand on top o f it dit not.

Vegetation succession

Initially open herbaceous vegetation was dom inant, m arked by wetter conditions at first (Subzone W a1), than by drier circumstances (Subzone W a2), com bined with greater aeolian activity. This is succeeded by the extension o f dune scrub (Juniperus), immediately followed by the increase o f influence o f w oody patches in the imm ediate vicinity (respectively Subzones W b 1 and W b2). The expansion o f w oodland and dune scrub happened in wetter circumstances as indicated by the relative great proportion of the Cyperaceae and the Aquatic pollen. Finally open vegetation becam e dom inant again (Zone Wc), w hat can be accredited to greater aeolian activity (severe drop o f Cyperaceae and Aquatic pollen, a new peak in the curve o f Asteraceae liguliflorae).

H tuna it influence on vegetation

A possible indication o f hum an influence is the presence - albeit in small quantities - o f Cerealia and Plantago lanceolata pollen. Cerealia w ere found in the upper'and low er levels. The probable presence o f Elymus arenarius , a natural com ponent of dune vegetation w hich produces also pollen o f Cerealia-type, makes it impossible to determ ine w hether or not there was cultivation of cereals. The curve o f Plantago lanceolata is quite continuous. Here again this species is considered also as specific for natural dune pastures, but w ether these have been affected by grazing is open to discussion (Jelgersma et at., 1970). Furtherm ore, Plantago pollen was in m any cases severely corroded, w hat m ade determination up to the species very difficult to achieve or even impossible (Plantago species).

Chenopodiaceae are considered to be indicators o f saline conditons (salt marshes), especially w hen this pollen-type occurs together with rem ains of m arine organisms such as cysts o f dinoflagellates (Straka, 1970). This is not the case here. Chenopodiaceae pollen can also point to influence o f m an on vegetation w hen it occurs together with other indicators o f such an influence (M unaut, 1967). In this w ay the com bined occurrence o f Cerealia, Plantago lanceolata, Chenopodiaceae and even Cruciferae could point tow ards influence o f man.

The presence o f Castanea pollen-type in the levels - 0 .2 7 and - 0 ,3 0 m is problematic. If the pollen o f this type belongs in fact to Castanea, then this could

THE IR ON A GE SETTLEM ENT A T DE PA NNE- WESTHOEK 33

indicate hum an influence, since Castanea was introduced by man in Rom an times (Godwin, 1956). Rom an influence in the upper levels is therefore not excluded. This interpretation is not contrary to absolute dating, w hen one reckons with absolute and relative errors in computing absolute dates. One could also consider possible infiltration o f Castanea pollen (11/18 mu), although M unaut (1967) estimates differential infiltration of this pollen-type to be improbable.

Taking into account the previous consideration, m an 's influence on the vegetation was very insignificant as no distinct signs were found, unless m ajor changes in vegetation w ould also be induced by m an such as the variations o f the content o f Asteraceae liguliflorae. Possibly Castanea w as introduced during Rom an times, w hat eventually affirms Rom an presence in the vicinity during a certain period.

G e o l o g i c a l a n d g e o m o r p h o l o g i c a l c o n t e x t (R.D.C.)

F. D epuydt (1966 and 1972) was the first to pu t forw ard an elaborate hypothetic evolution o f the dune area founded on geomorphological and sedimentological evidence. He assum es the presence of a fossil beach half-way underneath the present Y ounger Dunes, This statement is based on a division o f the W esthoek area into tw o distinctive zones (Fig. 6) : a southern zone (II) formed in the period 800 till 1400 A.D. and built up from a fossil beach beneath the northern zone (III) ; zone III is form ed after 1400 A.D. and built up from the present-day beach. The author how ever did not take into account the archaeological evidence. In fact extensive excavations proved the existence o f a vast archaeological site o f Iron and Rom an Age icfr. supra) in situ in the northern zone ! N o beach deposits have been found on top o f this site (Rahir, 1930, 76).

A D I N K E R K EDE PANNE

1000m_ J

0L

F ig . 6. - Evolution o f the dune area, according to Depuydt, 1 9 6 7 :

I. Dune zone with adjacent beach formed during the Subboreal age.II. Dune zone with adjacent beach formed between 800 (post D-2) and 1400 A.D.

III. Dune zone with adjacent present-day beach formed between 1400 A.D. and present.

34 R . DEC EUNYNCK AND H. THOEN

This contradiction between the archaeological evidence and the geomorphologi­cal interpretation was noticed by J. Herbauts (1971). This author tried to solve it by suggesting an older age for zones I I and III : respectively pre-Dunkerque-2 age and post-Dunkerque-2 age. In this w ay he still accepted the presence o f a fossil beach on top o f the site, contrary to the description of Rahir (1930, 76). Independently o f J. Herbauts, H. Thoen proposed another solution, in the context o f a study o f the Belgian Coastal area in the Rom an period (Thoen, 1978). He supposes that the site was sheltered by Older D une ridges, one landw ards of the site (II) and one seawards of the site (III) (Fig. 7). He also assumes a similar development as determined in The N etherlands (Jelgersma et al., 1970).

A D I N KERKEDE PANNE

nr

1000m I0LY* V4 532

F ig . 7. - Section through the dune area, according to Thoen, 1978 :

1. Older Dune ridges (I, II, III).2. Younger Dunes (IV).3. Marine deposits of Dunkerque-2 age.4. Peaty deposits.5. Habitation layers.

DE PANNE

o o o

0 500 m1 I I I . I

F ig . 8. - Geological section, after Lebbe and De Ceunynck, 1980 :

1. Older marine deposits.2. Older dune sediments.3. Subrecent beach deposits.4. Marine clay deposits probably at least partially of Dunkerque-2 age.5. Younger Dunes.

THE IR ON A GE SETTL EM EN T A T DE PA NNE- WESTHOEK 35

M eanw hile data from new borings are available (Lebbe and De Ceunynck, 1980). They give a better look on the spread of the older dune sediments beneath the Younger Dunes (Fig. 8). It appears that these sedim ents w ere eroded at their northern side by m arine action. The southern boarder of these sedim ents runs approxim ately half-w ay the area (Fig. 1 and 8). N o beach deposits w ere found on top o f the older dune sedim ents except for their northern boarder. The Older Dunes w ere probably levelled at the beginning of the Younger D une form ation. Furtherm ore dune deposits o f R om an age are included in the older dune sediments. The distribution o f those different older dune sedim ents is not yet clearly know n. This makes a reconstruction o f Iron Age dune landscape impossible, although the Older Dunes w ere then m ost likely extended m ore seawards.

*

The hum ic layer is one o f the hum ic and peaty strata o f the older dune sediments. A t the find-spot the overburden consists o f younger dune sediments. The upper level of the hum ic layer show s influence o f aeolian activity at about 1965 ± 110 B.P. (cfr. supra). From this level on there is a total rup ture in the deposits w itnessed by the different sedim entary structures, a very drastic increase o f the C aC 03 content, the total absence of pollen and the presence o f disturbed m ainly Rom an pottery fragments. This points definitely to later erosion at the top of the layer, in the course o f w hich a certain am ount o f sedim ent was rem oved and replaced by a younger sedim ent w ith typical aeolian character. This aeolian activity has nothing to do w ith the form ation o f the older dune sedim ents and the aeolian activity witnessed by the pollen content at the top of the layer. As a consequence it is highly probable tha t a younger vegetation o r occupation level (of R om an age) lay on top of the hum ic layer. It is as yet impossible to determ ine w hether this layer and the presum ed younger one w ere separated by a th in band o f aeolian sand as was sometim es indicated (de Loë, 1908, 37-38 ; Loppens, 1932, 95), although the pollen content at the top how ever points in that direction.

The hum ic layer w as at the onset possibly form ed as a soil in a dune depression, w hich subsequently becam e occupied by m an, w ithout how ever changing severely the natural vegetation in the im m ediate vicinity (cfr. supra). This vegetation is reflected in the diagram. Following Jelgersm a et a!. (1970) the phases o f soil form ation represent w et conditions, those o f dune form ation drier ones. In o u r case too, conditions seemed to be relatively wet. The same authors also observed a cyclicity in dune form ation alm ost synchronuous w ith the alternation o f transgressive and regressive phases in the coastal area : the w et phases coincided w ith the transgressive, the dry phases w ith the regressive phases. The dating o f the hum ic layer (top : 1965 ± 110 B.P., Hv-9136 ; bottom : 2660 ± 100 B.P., H v -9137) corresponds very closely w ith the archaeological m aterial and w ith the D unkerque-1 transgressive phase.

Tw o conclusions can be draw n : first a dense vegetation apparently developped in the nearby dunes during relatively w et conditions and secondly this

36 R . DE CEUNYNCK AND H. THOEN

developm ent coincides w ith the D unkerque-1 transgressive phase. It therefore appears that the same cyclicity applies at least partly for the dune form ation in Belgium.

T h e A r c h a e o l o g i c a l F i n d s (H.T.)

W e shall discuss only the archaeological finds recovered in situ , i.e. found within the hum ic layer. They consist exclusively o f pieces o f earthenw are, some 50 in total, which is rather a high num ber taking into account the limited cuttings. The distribution am ongst the three cuttings is uneven. M ost o f the finds come from section n° 1, the rem ainder from sections n° 2 and 3. In sections 1 and 2 most of the finds were deposited in the lower part o f the hum ic layer, the upper horizons containing few or no material at all. Section n° 3 contained only some very small fragments in the upper part. It should be noted that the ceramic material in the upper part consisted exclusively o f fragments o f briquetage. M ost o f these were too friable to be recovered.

The ceramic material can be divided in potsherds and briquetage.

I. The Pottery (Fig. 9 and 11)

N ine sherds belonging to dark-coloured, thick-walled hand-m ade earthenw ere pots. The past is greyish-black to black and fairly coarsely tempered, m ainly with

Fig. 9. - De Panne-Westhoek. Pottery from humic layer. (Photo H. Thoen, S.A.G.).

THE IR ON A GE SETTLEMENT A T DE PA NNE- WES THOEK 37

grog and small lum ps o f baked clay ; traces o f organic material are often visible. The walls are roughly smoothened. The outer surface is baked beige-brown, the inner beige-brow n for the th inner sherds and greyish-black for the thicker ones (max. wall thickness is 11 mm.). One sherd has on the outside a rather crude groove decoration (fig. 9,3). Three m atching sherds form part o f a rim of a beige- b row n bow l w ith a slightly carinated profile and a thickened rim -lip (fig. 9,4 ; 1 1 , 1 ).

As to the dating o f this pottery one can only notice the characteristics o f the pre- Rom an Iron Age earthenw are ; further chronological data cannot be obtained from so scarce a material.

f ' i t ■ ! ■ ¡ I ; ; ; ■ ; . i

I S I S B B B ■ H B i H

Fig. 10. - De Panne-Westhoek. Briquetage. (Photo H. Thoen, S.A.G.).

II. The Briquetage (Fig. 10-11)

Some forty fragm ents o f earthenw are show an even pale orange-brow n coulour (M unsell 7.5 YR 7 .5 /5 ) and are tem pered m ainly w ith organic material. They are very friable and show num erous cracks, clearly due to exposure from high tem perature. This material is know n in the archaeological literature as briquetage (cfr. introduction). This term denotes the technical im plem ents used in a salt- m aking process, w hereby a concentrated saline solution (brine) was evaporated in earthenw are recipients. For the different systems w e refer to the general studies by

38 R . DE CEUN YNCK AND H. THOEN

N enquin (1961), Riehm (1961), G ouletquer (1970), de Brisay and Evans (1975), etc., which treat these various systems in detail. Suffice it to say here that the briquetage technique is an universal process of salt-m aking which was in use in pre-historic Japan (Kondo, 1975) as well as nowadays in some parts of Africa (Gouletquer, 1975). In Europe, briquetage as a salt-making technique was in use from the Neolithic period till in the Middel Ages, W ell-know n centres existed in Central Europe (Poland, Salzkammergut, Saale Valley), in France (Lorraine and Brittany) and in England (Red Hills) (Gouletquer, 1971 and 1974).

In 1975 we published a short survey o f the ancient salt-making sites in Belgium. A part from the tw o Iron Age briquetage sites at De Panne and Brugge, we m entioned also the w ooden constructions at Zeebrugge and Raversijde (near Oostende) w hich w ere used as “marais salants” and date from the Rom an Period. Since then we have been able to examine a Rom an “Red HilP'-site at Leffinge, also in the vicinity of Oostende. Here too, briquetage was in use (excavations by V.O.B.O.W., publication forthcoming).

Despite the fragm entary nature o f the briquetage finds from the hum ic layer at De Panne, there are various elements recognizable. Tw o fragm ents doubtlessly belong to "clay nails” (de Loë and Rahir) (Fig. 10,14-15 ; 11,2-3). Both are square in section, one having a side o f 2.8 cm and a length o f 4.5 cm, the other a side o f 3 cm and a length o f 5.4 cm. The function o f these objects is not clear (pedestals ? grate ?). A nother fragm ent is a pinched clay prop (Fig. 11,4). These objects are not uncom m on am ongst the briquetage material and w ere used to prop up or connect

Fig . 11. - De Panne-Westhoek. Ceramic material from humic layer : 1 pottery ; 2-6 briquetage. Scale 1/3.

THE IRON A GE SETTLEM ENT A T D E PANNE- WESTHOEK 39

parts o f a grate or clay m oulds. In the latter case G ouletquer (1970, 84, fig. 15.4) speaks o f “houlettes de calage". The m ajority of the briquetage finds are thin sherds (3.5-5 mm), w hich com e from the clay m oulds w hich w ere destroyed on recovery o f the salt-cakes. T w o base fragm ents suggests that these m oulds had a round shape, the diam eter o f the base being around 3 cm (Fig. 10,12-13 ; 11,5-6).

The im portance o f these finds is m ainly the attestation of the use o f the briquetage technique in the Iron Age site at De Panne-W esthoek. The fact that this small am ount o f fragm entary m aterial enables us to gain new and im portant data on the salt-m aking technique in use at this site pleads for a total revision o f the finds from De Panne, w hich are still present in the reserves of the M usées Royaux in Brussels.

C o nclusion (R .D .C . & H . T .)

According to the radiocarbon m easurem ents the hum ic layer discovered át De Panne-W esthoek was form ed betw een 2660 ± 100 B.P. (710 b.c.) and 1965 ± 1 1 0 B.P. (15 b.c.) (conventional C-14 years). This concurs w ith the archaeological data and coincides in geological term s w ith the D unkerque-1 transgression.

The form ation o f the hum ic layer is the result o f the com bination o f soil- form ing processes - m ostly at the onset - and o f hum an activity. The paly- nological exam ination pointed out that at the outset an open herbaceous vegetation w as predom inant, initially in hum id, thereafter in drier conditions probably associated w ith a greater aeolian activity. The m ain expansion o f the vegetation started with the developm ent o f a dune scrub (Juniperus) and was followed by an increased influence o f strips o f dune-oak forest in the im m ediate vicinity, probably under hum id circum stances. The top o f the hum ic layer show s a strong decline o f scrub and forest vegetation, bu t a sudden rise o f herbaceous vegetation, probably as the result o f renew ed aeolian activity in drier circumstances.

The archaeological finds recovered during the sam pling o f the soil profiles consist of som e Iron Age pottery sherds and some fragm ents of briquetage , i.e. technical im plem ents used in salt-making. Judging from the latest archaeological literature, the pre-R om an site at De Panne-W esthoek should be dated back to the Late Iron Age, i.e. the La Tène period. On the spot w here the samples w ere taken, the m ajority o f the archaeological rem ains w ere found in the low er part o f the layer, the upper part yielding only a few scattered fragm ents. This suggests a vivid hum an activity during the initial form ation o f the hum ic layer and a m uch lesser hum an influence during the final phase. A lthough the radiocarbon date 2660 ± 100 B.P. is related to the initial form ation o f the hum ic layer, rather than to the start o f the hum an activity, w e do not exclude that the earliest hum an presence in the dunes o f De Panne m ight be older than hitherto accepted, and therefore m ight date back to the E arly Iron Age (Late Hallstatt).

The pollen diagram does not provide us w ith clear indications o f the hum an influence on the vegetation. A part from possible hints at dune pastures, specific

40 R . DE CEUNYNCK AND H. THOEN

culture-pollen are few o r non-existent. H um an presence has thus clearly been o f negligible effect on the dune vegetation. The m ain activity o f the com m unity will have been salt-making, as attested by the archaeological evidence.

C ontrary to earlier conceptions, the settlem ent did not come into being on an old beach flat bu t in the old dunes, w hich protected the site in the N. against the sea. On the other hand nothing points to a sim ilar protective barrier in the S. : the younger dunes here rest directly on m arine sediments. It should be noted how ever that no such sedim ents have been found on the site itself. The geological data and the archaeological rem ains indicate that all archaeological find-places (Panne I, II and III) are in fact m odern clearings o f one large archaeological site w hich extends below the present topographical features.

As to the final phase o f the pre-Rom an settlem ent and the continuity thereof into the G allo-Rom an period, w e m ay point out tha t - at least on the sam pling spot— the boundery layer betw een the hum ic layer and the covering dune sand show s clear signs o f erosion. It is therefore possible that the upper part o f the hum ic layer has been eroded in later aeolian phases. H ow ever, it should be noted that the upper zone o f the hum ic layer show s a sudden change from scrub and forest to herbaceous vegetation. M ost probably this indicates aeolian activity w hich w ould have been responsible, around 1965 B.P., for stopping further grow th of the hum ic layer. Judging from the very sporadic presence o f archaeological material in the upper horizon o f this layer, hum an activity in this period m ust have been very low. Som e Castanea-pollen in this horizon m ight indicate the first - if still sporadic - Gallo-Rom ^n presence in the vicinity.

A c k n o w l e d g e m e n t s

The authors wish to express their sincerest thanks to a ll who have m ade possible this s tu dy :

— the “Bestuur van W aters en Bossen" an d Ir. J. De Schuyler, fo r perm ission to investigate th e geology o f the dune area o f D e W esthoek ;

— the “Instituut tot aanm oediging van het W etenschappelijk Onderzoek in Nijverheid en L andbouw (I.W .O.N.L.)", fo r providin g a p ost-g radu a te gran t to R . De Ceunynck, which enabled him to in vestigate the geology o f the coasta l dunes ;

— Prof. Dr. G. De Moor, fo r prom oting the research o f R. D e C eunynck ;— Dr. V. de G roote a n d Dr. C. Verbruggen, fo r their valuable contribution to the

palynological research ;— D r. C. B aeiem an (Belgische G eologische Dienst), for her help in describing an d sam pling

th e profiles ;— Dr. M. A. Geyh (N iedersächsisches Landesam t für Bodenforschung, Hannover), fo r the

radiocarbon datings ;— Prof. Dr. J. de Heinzelin, Dr. L. Lebbe, Dr. A. G autier, F. De R aeve, PI. Andries, J.

Bourgeois an d m any others fo r their help in m any ways.

The draw ings w ere m ade by R , D e C eunynck (Fig. I), M. F lebaut (Fig. 2 - 8 ) an d H. Thoen (Fig. 11) ; the photographs by H. Thoen (Fig. 9-10). The archaeological p a r t o f the tex t was tran sla ted by E. Cools.

THE IRON AGE SETTLEMENT A T DE PANNE- WESTHOEK 41

B ib l io g r a p h y

B o e r b o o m , J. H. A. & Z a g w i j n , W . H., 1966. Pollenanalytical investigations in the coastal dune area near The Hague, The Netherlands, Acta Botanica N eerlandica, 15, 376- 388.

C h r i s t i a e n s , L., 1976. Lucht fotografische stu die van d e evolutie van de kust ten westen van D e Panne en tussen Bredene en W enduine. Dissertation Ghent U niversity (un­published).

C u m o n t , G., 1907. M onnaies trouvées dans les gisem ents côtiers de La Panne (Flandre Occidentale), Annales de la Société' d'Archéologie de B ruxelles, 21, 73-79.

d e B r i s a y , K . W . and E v a n s , K . A . (ed.), 1975. Salt. The S tu dy o f an ancient Industry. Report on the Salt W eekend held at the University o f Essex 20, 21, 22 September1974. Colchester.

D e L a e t , S. J., 1974. Prehistorische Kulturen in het Zuiden der Lage Landen. U niversa, W etteren.

D e L a e t , S. J., 1979. Prehistorische Kulturen in hei Zuiden der L age Landen. 2nd. ed.. U niversa, W etteren.

d e L o ë , A. 1901-02. La Station préhistorique, belgo-rom aine et franque de La Panne, com m une d ’Adinkerke (Flandre Occidentale), note préliminaire, Bulletin de la S ociété royale belge d ’AnthropoIogie et d e Préhistoire, 20. 1-8.

d e Loë, A ., 1908. Continuation des Fouilles de La Panne, Bulletin des M usées royaux d ’A rt et d ’H isto ire, 1908, 35-39.

d e M a e r e d 'A e r t y c k e , Bon, 1905. La collection d'objets anciens de La Panne déposée à Gruuthuse, Annales de la S ocié té d'Em ulation de Bruges, 55, 36-46.

D e p u y d t , F., 1966. A nalyse van de Strand- en Duinsedim enten in de Belgische W esthoek, A cia Geographica Lovaniensia, 4, 68-82.

D e p u y d t , F . , 1972. De Belgische S tran d- en Duinform aties in hel kader van de Geo- morfologie der zu idoostelijke N oordzeekust. Verhandelingen van de Koninklijke A ca­dem ie voor W etenschappen, Letteren en Schone Kunsten van België - Klasse der W etenschappen, X X X IV , N r. 122. Brussel.

D o i n g , H . , 1974. Landschapsoecologie van de duinstreek tussen W assenaar en IJnniiden. M ededelingen van de L andbouw hogeschool te W ageningen, 74-12. W ageningen.

D o n n y , G., 1886. Sur l'existence de vestiges d u n établissem ent gallo-rom ain dans les dunes de La Panne, Bulletins de l ’A cadém ie royale des Sciences, des Lettres e t des B eau x- A rts de Belgique. 56, 559-561.

F a v o r e l , D om P., 1961-62. La Région de La Panne à l’époque prérom aine, rom aine et franque, Handelingen van de Koninklijke G eschied- en O udheidkundige K ring van K ortrijk , 32, 4-223.

G o d w i n , H., 1956. The H istory o f the British Flora. Cambridge U niversity Press.G o u l e t q u e r , P. L ., 1970. Les B riquetages Arm oricains. Technologie protohistorique du sei

en A rm orique. Travaux du Laboratoire d ’AnthropoIogie préhistorique, Série C 105- 22. Rennes.

G o u l e t q u e r , P. L„ 1974. The D evelopm ent o f Salt-M aking in Prehistoric Europe. R e­printed from E ssex Journal, 9, N ° 1.

G o u l e t q u e r , P. L.. 1975. Niger, C ountry o f S a lt, in : d e B r i s k a y , K. W . and E v a n s , K. A .,1975, 47-51.

42 R . DE CEUN YNCK AND H. THOEN

H e r b a u t s , J. 1971. Flore et végétation des dunes de la R éserve N aturelle dom aniale du W esthoek. Bestuur van W aters en Bossen, W erken, nr. 5. Brussel.

J e l g e r s m a , S . , d e J o n g , J . , Z a g w i j n , W . H . and v a n R e g t e r e n A l t e n a , J . F . , 1970. The coastal dunes o f the W estern Netherlands -, geology, vegetational history and archaeology, M ededelingen R ijks Geologische D ienst, N .R ., nr. 21, 93-167.

K a l k h o v e n , J. T. R., S t u m p e l , A. H. P. and S t u m p e l - R j e k s , S. E., 1976. Landelijke M ilieu­kartering. Een beeld van de n atuur in N ederland. Rijksinstituut voor Natuurbeheer, Verhandeling nr. 9. 's-Gravenhage.

K o n d o , Y., 1975. The S a lt Industry in Ancient Japan, in : d e B r i s a y , K. W . and E v a n s , K. A ., 1975, 61-65.

L o p p e n s , K., 1928. Sur quelques fouilles faites dans les dunes de La Panne, Bulletin de la S ocié té royale belge d ’AnthropoIogie et de Préhistoire, 43, 232-239.

L e b b e , L. and D e C e u n y n c k , R., 1980. Lithostratigrafie va r het duingebied ten W esten van D e Panne. Professional Papers. Belgische G eologische Dienst. Brussel (publication forthcoming).

L o p p e n s , K. 1932. L a Région des Dunes de Calais à Knocke. Coxyde.M a e r t e n s ( d e N o o r d h o u t ) , J., 1903. Fouilles à la station préhistorique de La Panne,

Bidletijn der M aatschappij van Geschied- en O udheidkunde te Gent, 156-162.M a r ië n , M . E ., 1952 . Oud-België. Van d e eerste landbouw ers tol de kom st van C aesar.

Antw erpen.M a r i ë n , M. E., 1971. Tribes an d archaeological groupings o f the La Tène period in

Belgium : som e observations, in : B o a r d m a n , J., B r o w n , M. A. and P o w e l l , T. G. E. (ed.), The European Com m unity in L ater Prehistory. S tudies in honour o f C. F. C. H aw kes, London, 211-241.

M a r ië n , M . E ., 1980 . Belgica A ntiqua. De stem pel van Rom e. Antwerpen.M o o r e , P. D. and W e b b , !l. A ., 1978. An Illustrated G uide to Pollen A nalysis. London.M u n a u t , A. V ., 1967. R echerches Paléo-Ecologiques en B asse et M oyenne Belgique. Acta

Geographica Lovaniensia, 6. Leuven.N e n q u i n , J., J96 1 . Salt. A S tu dy in Econom ie Prehistory. D issertationes Archaeologicae

G andenses, VI. Brugge.R a h i r , E., 1927, L’A ge du Fer à La Panne. U n e Fabrique de Poteries, Bulletin de la Société

royale belge d ’AnthropoIogie et d e Préhistoire, 42 , 14-67.R a h i r , E., 1928. Vingt-cinq années de Recherches, de R estaurations e t d e Reconstitutions.

Bruxelles.R a h i r , E., 1930. La Panne. Fabrication de Poteries. Habitats et Sépultures de l'Age du Fer.

R om anisation des Habitants de l’A ge du Fer, Bulletin de ¡a Socié té royale belge d'Anthropologie e t de Préhistoire, 45, 10-82.

R ie h m , K., 1961. Prehistorie Salt-Boiling, A n tiqu ity , 35, 181-191.R o e l e v e l d , W ., 1976. The Holocene Evolution o f the Groningen M arine-C lay D istrict. Ber.

Rijksdienst v oor Oudheidk. Bodem onderzoek, 24, 1974, supplem ent. ’s-Gravenhage.S t r a k a , H., 1970. Pollenanalyse und Vegetationsgeschichte. W ittenberg Lutherstadt.T h o e n , H., 1975. Iron A ge an d Rom an sa lt-m ak in g S ites on the Belgian C oast, in : d e

B r i s a y , K. W . and E v a n s , K. A ., 1975, 56-60.T h o e n , H., 1978. De Belgische K u stv lak te in de R om einse tijd. B ijdrage tot de stu die van de

landelijke bewoningsgeschiedenis. Verhandelingen van de Koninklijke A cadem ie voor W etenschappen, Letteren en Schone Kunsten van België - Klasse der Letteren, XL, Nr. 88. Brussel.