The Mont-des-Récollets section (N France): a key site for the Ypresian-Lutetian transition at mid-latitudes – reassessment of the boundary criterion for the base-Lutetian GSSP

ABSTRACT Re-excavation of the famous Mont-des-Récollets quarry in northern France and re-interpretation of the adjacent Cassel borehole have led to the reconstruction of one of the most complete Upper Ypresian and Lutetian stratigraphic successions of the southern North Sea Basin. It includes the entire suite of formations and members, from the top of the Hyon Sand Formation (top NP12) to the base of the Maldegem Formation (mid-NP15), originally defined in central Belgium, extending the lateral distribution of most of these units up to northern France. The similarity in lithofacies and calcareous nannofossil assemblages and the identification of in situ specimens of Nummulites laevigatus (Bruguière, 1792) and Campanile giganteum (Lamarck, 1804) at the Mont-des-Récollets bear witness to direct north-south connections between the Belgian Basin and the Paris Basin during Biochron NP14 and early Biochron NP15. However, direct connections have only been intermittent. During late Biochron NP12 and Biochron NP13, both basins became disconnected because of lowering of the global sea level in combination with uplift of the Paris Basin. The massively reworked large-sized N. laevigatus and the presence of N. laevigatus-bearing sandstone blocks and lignitic pebbles at the base of the Lede Formation indicate a major erosion phase at the NP14-NP15 Biochron transition in the Belgian Basin. This was probably due to uplift of the Brabant Massif. The new Mont-des-Récollets data, including the introduction of a new genus (Luminocanthus n. gen.) and ten new species (Blackites minusculus n. sp., Blackites praeinflatus n. sp., Luminocanthus eolutetiensis n. gen., n. sp., Luminocanthus plenilutetiensis n. gen., n. sp., Martiniaster cecellanoriae n. sp., Nannoturba joceliniae n. sp., Sphenolithus quadricornutus n. sp., Sphenolithus recolletensis n. sp., Trochoaster nodosus n. sp. and Trochastrites pyramidalis n. sp.) have allowed to substantially refine the calcareous nannofossil stratigraphy. The subdivision of Zone NP13 into 3 subzones (NP13-a to NP13-c) as well as the major nannofossil turnover at the base of Zone NP14 and the nine nannofossil-events within Zone NP14 have been recognised throughout the Belgian Basin. Several of these have been recorded in the Paris Basin, the Hampshire Basin and the Aquitaine Basin, highlighting their interbasinal correlation potential. The nannofossil data provide for the first time evidence that the base of the ‘Chaumont-en-Vexin sands' (term informally introduced herein to designate the base of the historical Lutetian stratotype) and the base of Unit A4 in the upper Aalter Sand Fomation are coeval or nearly coeval. This is also the case for the base of the overlying units, the ‘Glauconie Grossière s.s.' (as redefined in the Paris Basin by Blondeau in 1980) and the Brussel Sand Formation (‘Bruxellian' as originally defined in Belgium by Dumont in 1839). This investigation and its continuation, detailed in a forthcoming study of the Brussel Sand Formation, also revealed that the internationally accepted base-Lutetian boundary criterion (lowest occurrence or LO of Blackites inflatus) is difficult to apply in the North Sea Basin because of the extreme rarity of this marker species (e.g. not recorded at the Mont-des-Récollets). Worst of all, its strict application would imply that the major part of the Brussel Sand Formation and both the ‘Chaumont-en-Vexin sands’ and the lower part of the overlying ‘Glauconie Grossière s.s.’, base of the historical Lutetian stratotype, should be of Ypresian age, which is a contradictio in terminis. To resolve this contradiction, it is suggested to amend the original criterion (LO of B. inflatus), proved to be inadequate at middle and high latitudes and diachronous at low latitudes, and to replace it by the lowest occurrence (LO) of Discoaster sublodoensis. The latter is part of a major calcareous nannofossil turnover (= BALCAT-event), which has been identified at the base of the historical Lutetian stratotype. This new proposal would mean that the base of the Lutetian should be lowered down in the Gorrondatxe GSSP, probably by about 130 m (c. 1.3 Myr), to a level within the middle of Chron C22n, around 800 kyr earlier than the LO of Turborotalia frontosa. If this is the case, then the base would range in age between 49.11 Ma and 49.20 Ma, depending on the age model used.


INTRODUCTION
The Mont-des-Récollets or Récollets Hill (highest point at 159 m above sea-level: 50°48'02.74"N, 2°30'23.06"E; map-sheet: XXIII-Cassel; x = 62.000, y = 344.500; z = 159.000) and its nearby twin Cassel Hill constitute the western end of an east-west oriented range of hills, extending through central Belgium and northern France (Fig. 1). These Flemish Hills roughly subdivide the Flemish Plain into a northern Dutch-speaking sector and a southern French-speaking sector. Their tops, which rise about 100 m above the surrounding plain, are formed of lower Eocene glauconitic sands and middle Eocene calcareous sands and clays. Ferruginous sandstones, which were recently assigned to the Flemish Hills Formation and for which a late Eocene age has been suggested (Houthuys 2014), make up the hills' caps. Their presence protected the hill-zone from post-Miocene erosion and denudation, which in the adjacent areas exceeded 100 m.
The stratigraphy of the Cassel area has been discussed since the dawn of geology (Desmyttere 1826; Elie de Beaumont 1833; d'Archiac 1839; Lyell 1852; amongst many others). Ortlieb & Chellonneix (1870) were the first to give a detailed description of the Mont-des-Récollets outcrop, identifying 27 beds in the main quarry ('La Grande Carrière'). Leriche (1921) carried out the most comprehensive and best-documented study of the geology of the Cassel area, mainly focusing on the Mont-des-Récollets outcrop (Fig. 2). In his remarkable monograph he described several sections on the Mont-des-Récollets, based on observations dating back to 1905, when the sands and clays were intensely quarried for industrial applications and brick-making. In the early 1920s the 'Grande Carrière' was already abandoned and soon afterwards progressively filled in. As a result, subsequent references to the stratigraphy of the 'Grande Carrière ', e.g. in Feugueur (1963), rely entirely upon the observations of Ortlieb & Chellonneix (1870) and Leriche (1921). However, additional information was gained through the study of the Western quarry at the Mont-des-Récollets (Leriche 1938) and the Cassel borehole (Blondeau et al. 1972) (Fig. 3). The different units in the Cassel borehole were dated on the basis of multidisciplinary microfossil investigations, including calcareous nannofossils, although with only poor resolution (Lezaud 1972).
In the middle of the 1980s the authors of the present study re-investigated the Mont-des-Récollets outcrop zone. The discovery of a small, 10 m-wide flat zone, surrounded by gentle hill slopes with dense vegetation and a few large trees, led to retracing the famous central quarry or 'Grande Carrière'. During the successive fieldwork campaigns the old quarry front was re-exposed after removing tons of quarry fill. A 14 m deep step-wise constructed pit was dug out from the top of the Lede Sand Formation down to the top of the Aalter Sand Formation (Figs 4;8;9). The front of the Western quarry, along the Dunkerque-Lille road, first described by Leriche in 1938Leriche in (1938, was freshened up with a bulldozer (Figs 5;6). Several shallow auger holes were drilled in the floor of both quarries in order to provide good overlaps between the quarries themselves and with a series of boreholes carried out at the 'Tir des Anglais' (= shooting-range of the English Army during the second World War), at about 300 m SE of the 'Grande Carrière'. These holes, drilled with water-injection, reached depths of about 16 m (Fig. 4). The Mont-des-Récollets sections were visited during the geological field trip of the 'Association des Géologues du Bassin de Paris' (AGBP) in May 1990. The geological outline of this field trip, including a location map and a simplified lithological draft of the Mont-des-Récollets sections, was published shortly thereafter (Nolf & Steurbaut 1990). The map and some of the logs were refigured in the excursion guide presented at the occasion of the 8 th International Palynological Congress at Aix-en-Provence, France (Schuler et al. 1992). Some of the figures were redrawn subsequently and included in the compilation work on the sequence stratigraphic interpretation of the Belgian Tertiary (Vandenberghe et al. Steurbaut E. & Nolf D. 1998). Fobe & Spiers (1992) and, later on, Hooyberghs (1999) used these documents to calibrate their respective lithological and foraminiferal data. During the last decade the Mont-des-Récollets outcrops have not been subject of further investigations. However, progress was made in the understanding of the stratigraphy of the different units encountered at Mont-des-Récollets: Gentbrugge Formation in the Zemst area (Steurbaut et al. 2015), Brussel Sand Formation in the Brussels area (Damblon & Steurbaut 2000) and Lede Sand Formation in the Brussels area (Herman et al. 2001) and in the Oosterzele-Balegem area (Smith et al. 2004). Interpretation and correlation of the Mont-des-Récollets strata were greatly advanced by the recent introduction of several new calcareous nannofossil taxa (Steurbaut 2011) and by the high resolution nannofossil studies of a series of ODP borehole sections in the Atlantic (Agnini et al. 2014) and of several outcrop sections in Belgium (Steurbaut et al. 2015) and in the Aquitaine Basin (Lin et al. 2017).
This study discusses the stratigraphical significance of the almost 100 m thick Mont-des-Récollets succession, believed to represent one of the most complete and most expanded upper Ypresian to middle Lutetian outcrop sections of the North Sea Basin (Fig. 1). It aims at clarifying the biostratigraphical and sequence stratigraphical context of the Ypresian-Lutetian boundary interval, through detailed analyses of lithological units and calcareous nannofossils. Consequently, it will represent one of the first mid-latitude test-cases for the identification of the base of the Lutetian Stage, outside its GSSP-defined stratigraphical context in Northern Spain (Gorrondatxe Section: Payros et al. 2009Payros et al. , 2012Molina et al. 2011) and other low latitude contemporaneous sections in Southeast Spain (Tori & Monechi 2013) and Northern Italy (Franceschi et al. 2015). The upper, about 20 m thick, post-Eocene or late Eocene (sensu Houthuys 2014) part of the section, which is badly exposed and heavily weathered, has not been studied in detail. GEODIVERSITAS • 2021 • 43 (11) Leriche (1921) up to the Late 1980's For his interpretation of the stratigraphical succession in the Cassel area, Leriche (1921) integrated the commonly accepted late 19th century views and concepts on Belgian Tertiary stratigraphy with his own paleontologically driven new insights in NW European geology. He grouped the different strata into six major categories or stages, in ascending order the Ypresian, the Bruxellian, the Ledian, the Bartonian, and last but not least the Diestian and Amstelian, both incorporated into the Pliocene (Fig. 3). Feugueur (1963) and Blondeau et al. (1972) essentially adopted Leriche's stratigraphic framework in their study of the Mont-des-Récollets outcrop and of the Cassel borehole respectively. However, the lower part of Leriche's Bruxellian, the glauconitic sands, were incorporated into the upper Ypresian by these authors and termed 'Zone à Turritelles et à Cardita planicosta' (Feugueur 1963) or Aalter Sands (Blondeau et al. 1972).

PREVIOUS LITHOSTRATIGRAPHIC INTERPRETATIONS OF THE MONT-DES-RÉCOLLETS SECTION
Leriche's detailed observations resulted in the introduction of a detailed chronostratigraphic framework, the general structure (units and limits) of which is more or less valid today (Fig. 3). The terminology, however, underwent substantial modifications. Through the years most of the stage names used by Leriche have become obsolete (Bruxellian, Ledian;see De Geyter et al. 2006 for a review) or were progressively reconverted into a new lithostratigraphic terminology (see material and methods). NoLf & steurbaut (1990) In their summary on the stratigraphy of the Mont-des-Récollets, presented on the occasion of the AGBP geological field trip, Nolf & Steurbaut (1990) identified the following units (in descending order): the "Diestian", resting on the Ursel Member, the Asse Member and the 'Bande Noire', which were grouped into the Kallo Formation. The latter overlies a series of formations, including, in descending order, the Lede Formation, the Brussel Formation, the Aalter Formation, the Vlierzele Formation and the Ieper Formation. No further stratigraphic details were given by these authors, except for the location of the stone layers and some conspicuous fossiliferous levels.

MATERIAL AND METHODS
The Mont-des-Récollets section is a composite of several outcrop and borehole sections, ranging from the top of the hill at 159 m above mean sea level standard N.G.F. ('le nivellement general de la France') to about 63 m N.G.F. The absolute height indications are obtained through precise levelling starting from the top of the hill at 159 m. All the values mentioned here are calibrated against this N.G.F. standard. For reasons of readability the notion 'N.G.F.' will not be endlessly repeated throughout the text. The oldest sediments were penetrated in a borehole on the southeast flank of the Mont-des-Récollets (at the 'Tir des Anglais'). Due to the drilling methodology (flushing), only small sandy clay chips could be recovered, together with some macrofossils and glauconite grains. Taking into account the exact position of the different sections (Fig. 4), it is clear that about 14 m section could not be investigated, mainly because of the drilling technique (shallow auger holes; the water-injection method could not be applied in the western quarry) and of the presence of a rather thick Quaternary cover or of slope debris on the southeast flank of the hill. Fortunately, analysis of the topographic situation and the lithological descriptions of the sediments in the nearby Cassel borehole, allowed the reconstruction of the entire succession. The lithostratigraphic nomenclature adopted here is from Steurbaut (2015), following earlier recommendations by Steurbaut (1998; first formal subdivision of the lower Ieper Group in formations and members), Laga et al. (2002; nomenclature of the Paleogene formations) and Steurbaut (2006; review of Ypresian stratigraphy). It differs from that in Nolf & Steurbaut (1990), essentially because of updating of the Ypresian terminology, including downgrading of the Vlierzele Formation to Member status and upgrading the Ieper Formation to group status. The Egem Sand Member, the Bois-la-Haut Sand Member and the Mont-Panisel Sand Member, subdivisions of the former Ieper Formation, were in 2015 regrouped into the Hyon Sand Formation (the latter was introduced by King 1994 andupdated in Steurbaut 2011   1974; a controversial term validated by Steurbaut & Nolf 1986), the Merelbeke Clay Member (De Moor & Germis 1971), the Pittem Clay Member (Geets 1979), the Vlierzele Sand Member (Kaasschieter 1961) and the Aalterbrugge Sand Member (Hacquaert 1939) are incorporated into the Gentbrugge Formation. This term replaces the previous term Gent Formation (Maréchal 1994;Wouters & Vandenberghe 1994), preoccupied for a unit within the Quaternary of Belgium (Laga et al. 2002: 140). An extended overview of the lithostratigraphy of the Ieper Group is given in Steurbaut et al. 2016a. The term Brussel Sand Formation, as used herein, corresponds to the classical Bruxellian concept of Dumont (1839), subsequently adopted by Leriche (1912Leriche ( , 1939. In its type-area this formation consists of poorly glauconitic whitish fine to medium-grained quartz sands, generally between 5 m to 10 m thick. Its thickness increases up to several tens of meters (? 80 m) in a 15 km wide trough between Brussel and Leuven (Houthuys 1990(Houthuys , 2011Damblon & Steurbaut 2000: fig. 5 (Wemmel, Schepdaal, Zemst, etc.) with thickness of about 9 m (cf. Steurbaut et al. 2015). In the centre of Gent (Blandijnberg or 'Colline de Saint-Pierre' auct.) it is reduced to about 1 m, including a basal, 5 to 10 cm thick, glauconite bed (Delvaux 1886;Hacquaert 1936;Leriche 1943a). A second somewhat thicker glauconite bed, which rapidly passes upward into glauconitic clays of the Asse Clay Member, overlies the Wemmel Sands in the Gent area.
The chronostratigraphic interpretation of the Mont-des-Récollets section has also undergone fundamental modifications since Leriche. The term Bruxellian (obsolete now) as used by Leriche (1921Leriche ( , 1943a corresponds in modern terminology to the combination of Hyon Sand Formation, Gentbrugge Formation, Aalter Sand Formation and Brussel Sand Formation, while the term Lede Sand Formation has replaced the ancient stage name Ledian. Calcareous nannofossil (Steurbaut 1986) and dinoflagellate cyst investigations (De Coninck 1995) also revealed that the 'Bande Noire' and the overlying Asse and Ursel Clays do not belong to the Bartonian, as thought by Leriche, but have to be included, together with the Brussel Sand Formation (or part of it, depending on the placement of the base of the Lutetian, see Fig. 3) and the Lede Sand Formation in the Lutetian.
Sixty samples were processed for calcareous nannofossil investigation following the preparation and investigation procedures explained by Steurbaut & King (1994) and Steurbaut & Sztrákos (2008), of which 39 proved to be productive. The taxonomy adopted in this work is essentially based on Perch-Nielsen (1985), taking into account the modifications by Wei & Wise (1989), Varol (1992, Young & Bown (1997), Aubry (1999), Bown (2005, Aubry & Bord (2009), Shamrock (2010b) and Steurbaut (2011). The new and poorly known taxa are discussed according to the alphabetic order of the families to which they belong.

Lithostratigraphic review of the MoNt-des-récoLLets sectioN
The Mont-des-Récollets is one of the few places in the southern North Sea Basin where the upper Ypresian to lower Lutetian Zenne Group is almost complete (Fig. 1). The lithostratigraphic units identified at this site and in the adjacent Cassel area are displayed in Fig. 3. The topographic and stratigraphic positions of the different subsections of the Mont-de-Récollets are outlined on Fig. 4. Detailed descriptions and photographs of the two main quarries, the lower Western quarry and the upper 'Grande Carrière' are given in Figures 6-7 and 8-9 respectively. The structure of the Western quarry, which is composed of several faulted blocks, is displayed on Figure 5. The stratigraphic units are discussed in ascending order.  (Steurbaut et al. 2015), has not been recognized as a separate unit in the Mont-des-Récollets area and in the adjacent Cassel Hill (outcrops and borehole). However, the available data are not sufficiently detailed to decide upon its presence in the Cassel borehole.
Merelbeke Clay Member. 3.10 m of grey stiff clay, tentatively attributed to the "Paniselian", has been penetrated in the Cassel borehole between 72.30 m and 75.40 m depth (Blondeau et al. 1972). Its lithological description as well as its stratigraphic position point to the Merelbeke Clay. The latter, widespread in the area between Gent and Brussels, has also been identified in the Rodeberg borehole, 18 km east of the Mont-des-Récollets (Gulinck, unpublished data from 1968) (see Fig. 1 for location). It should also be present at the Mont-des-Récollets, although not yet identified because of falling within the non-exposed interval (see Fig. 4, also above).
Pittem Clay Member. The sandy clay with basal glauconitic sand observed in the nearby Cassel borehole (between 72.30 m and 71.40 m) and included in the "Paniselian" by Blondeau et al. (1972: 23) is believed to represent the Pittem Clay. This interpretation is based on lithologic characteristics, supported by calcareous nannofossil dating (NP 13; interpretation based on Lezaud 1972). The Pittem Clay has not been recognised at the Mont-des-Récollets because of falling within the non-exposed interval.
Vlierzele Sand Member. Exposures of these heterogeneous glauconitic fine to medium-grained sands have been frequently observed in the area between the Kluisberg ( Fig. 1[point 7]), Gent and Brussels (Kaasschieter 1961;Steurbaut & Nolf 1986;Steurbaut 2006), although almost never in their full extent.  Fig. 13). In the Mont-des-Récollets area, the junction with the underlying Vlierzele Sand is slightly erosive and marked by a decrease in grain size and the income of thin undisturbed dark clay beds. The contact with the overlying Aalter Sand Formation is sharp.  The yellowish green glauconitic fine sands, rich in molluscs, exposed between 117.35 and 99.10 m in the lower part of the 'Grande Carrière' and in the Western quarry at the Mont-des-Récollets are attributed to the Aalter Sand Formation. This attribution is based on lithological similarity with the Aalter Sand stratotype (Steurbaut & Nolf 1989): presence of khakicoloured glauconitic fine sand, turritelline-beds, Venericor (previously Megacardita) planicosta beds, and some incompletely indurated calcareous sandstones. At the Mont-des-Récollets this 18.25 m thick Aalter Sand succession can be subdivided into four units. Detailed lithological desciptions of these units are given in Figures 7 and 9.

Aalter
Unit A1. It is represented by 4 m greyish bioturbated mediumgrained sand. The contact with the underlying Aalterbrugge Member is sharp and marked by a slight increase in grain size and in glauconite content. In the middle of the unit occurs an accumulation of V. planicosta, labelled (Ve) on Figs 5-7.
Unit A2. This unit, which is about 3 m thick, is bounded above and below by bioturbated surfaces. It differs from the underlying Unit A1 by the presence of almost monospecific shell beds, composed of either turritelline or V. planicosta shells.
Unit A3. It ranges from the bioturbated surface at 106.60 m to the base of glauconitic highly fossiliferous sand located at c. 114.20 m, at 1.5 m below the top of the Western quarry. The contact with the overlying unit could not be precisely specified (sharp, burrowed, etc.?) because it occurs beneath the quarry floor of the 'Grande Carrière', where it was only penetrated in the borehole, and was beyond reach in the Western quarry. However, there seems to be a substantial increase in glauconite at the junction (c. 114.60 ?Vli a

The Western quarry section at the Mont-des-Récollets
Unit A4. Sandstone concretions, rich in the bivalve Orthocardium porolosum, occur frequently within this unit, which is furthermore characterised by its texture (fine sand) and the presence of fine-grained glauconite and other bivalves (oysters). The contact with the overlying Brussel Formation is sharp and marked by an increase in grain size.  In the Mont-des-Récollets area the Brussel Sand Formation is very similar to what has been observed in the Brussel area (Burtin 1784; Galeotti 1837; Rutot 1882; Leriche 1912Leriche , 1923Gulinck 1963;Houthuys 1990). It consists of 5.65 m thick whitish fine to medium-grained quartz sand with shell debris and several beds of sandstone with typical "quartzitic" patina. On the basis of lithologic features it can be subdivided into four units (see Fig. 9).

Brussel
Unit B1 (117.35-118.95 m). This unit consists of uniform whitish fine to medium-grained sand with echinoid (among which Maretia omaliusi) and bryozoal debris and a few sandstone banks. Its boundary with the underlying Aalter Sand Formation is marked by a major increase in grain size and a substantial decrease in glauconite.
Unit B2 (118.95-122.15 m). There is a substantial break in sedimentation between Unit B1 and Unit B2, as shown by the interburrowed surface. Unit B2 differs from Unit B1 by a much stronger heterogeneity and a coarser grain-size, becoming extremely coarse at the base. Its lower part (Subunit B2a) includes two sandstone levels, which seem to be consistently occurring, and which are overlain by obliquely stratified medium-grained sand. Its upper part (Subunit B2b) essentially consists of whitish bioturbated, locally shelly, mediumgrained sand (Figs 8; 9).
Unit B3 (122.15-122.55 m). This unit, consisting of 20 cm thick white yellowish shelly fine sand, capped by a 20 cm thick bivalve-bearing sandstone yielding large nautiloid shells, is separated from the underlying unit by an interburrowed junction. It is furthermore marked by a finer grain size and the common occurrence of oyster shells, especially at the base.
Unit B4 (122.55-123.00 m). This white yellowish shelly fine sand with rusty stains is above all marked by the occurrence of common small-sized (5 to 8 mm in diameter) Nummulites laevigatus in situ (around 5 specimens/kg sediment). There is a sharp contact with the underlying sandstone. The "Grande Carrière" section at Mont-des-Récollets fossiliferous sandstone with up to 2 mm large quartz and flintstone particles, abundant bivalves, shark teeth and reworked N. laevigatus whitish sand with rusty stains, molluscs and frequent in situ N. laevigatus sandstone with "quartzitic" patina and dispersed bivalves white yellowish fine sand, highly fossiliferous (mainly oysters) at the base whitish medium to coarse-grained quartz sand, fining upward; bioturbated, dispersed mollusc accumulations and echinoid shell debris (Maretia omaliusi) white to greyish medium to coarse-grained sand with oblique stratification (between 18° and 25° dipping to the east); dispersed shell fragments white sandstone with "quartzitic" appearance white to greyish medium to coarse-grained sand, rich in quartz grains white sandstone, with rusty weathered upper surface very coarse quartz sand, rich in echinoid shell fragments whitish medium-grained sand with shell debris, bioturbated, glauconitic and micaceous thick sandstone bank, consisting of hugh massive blocks with a "quartzitic" appearance white, slightly greenish, fine to medium-grained sand, rich in echinoid and bryozoal fragments; local sandstone blocks greyish sandstone yellowish green fine sand, glauconitic, rich in oysters highly fossiliferous limestone with abundant Orthocardium porulosum, glauconitic yellowish green fine sand, with abundant Orthocardium porulosum glauconitic; local sandstone concretions sandstone rich in Orthocardium porulosum sand with frequent sandstone concretions, rich in molluscs greyish green sand, shelly, glauconitic, locally cemented fossiliferous clayey sand, highly glauconitic clayey sand, glauconitic, rich in shell fragments less fossiliferous locally cemented glauconitic fine sand with dispersed shells, locally cemented sandstone layer or hard sandstone concretion 1-8, sandstone levels numbered after Leriche (1921)  iferous fine sand and calcareous sandstone and limestone banks with particular characteristics (Figs 8; 9). The sandstone bank at 124.3 to 124.5 m (level 4) contains several large fragments of Campanile giganteum and represents one of the few records of this extremely large gastropod outside the Paris Basin (see Discussion). The overlying sandstone level 5 is known for the commonly occurring nautiloid shells. The Lede Sand Formation differs from overlying and underlying units by the presence of small Palaeonummulites variolarius. It is widespread in central and north Belgium, reaching a mean thickness of about 6 m. A gravel bed, consisting of rolled sandstone pebbles, shark teeth and other fossil debris, always marks its base. The latter presents a wide range of textures, sometimes loose (e.g. at Oosterzele, Smith et al. 2004), but often (semi-)lithified (e.g. at Balegem, Jacobs & Sevens 1994). At the Mont-des-Récollets it consists of large fossiliferous sandstone blocks, some extremely rich in large-sized N. laevigatus, which are locally underlain by a very thin rusty badly sorted fine sand with up to several mm thick quartz grains and abundant reworked large-sized N. laevigatus and other fossils (molluscs and a few shark teeth).
Maldegem  At the Mont-des-Récollets the base of the Maldegem Formation consists of an appoximately 1 m thick heterogeneous unit, including a basal 25 cm thick pure glauconite with some rare shell debris and Palaeonummulites variolarius, 40 cm of clay, a 30 cm thick Lentipecten-shell bed and glauconitic clayey sands ( Fig. 9), very similar to what has been observed in the centre of Gent (Blandijnberg or 'Colline de Saint-Pierre' auct., see Material and Methods). Lyell (1852: 327) introduced the term 'Bande Noire' for the basal glauconite bed situated in a similar position in a pit on the eastern slope of the Cassel Hill. The above mentioned 1 m thick heterogeneous complex, also recorded in the Cassel borehole, is considered to represent the Wemmel Sand Member. As only the lower half a meter yielded calcareous nannofossils it is not clear up to now if this unit only represents the lower part of the member, the top of which has been eroded, or if it corresponds to the complete, but very condensed, Wemmel Sand Member. This heterogenous unit is overlain by very clayey sands and glauconitic sandy clays, which are not studied here. In the Cassel borehole (Blondeau et al. 1972) this predominantly clayey succession (clay content around 40%) is about 8 m thick, and believed to represent the Asse Clay Member, The overlying 6 m thick non-glauconitic grey stiff clay is attributed to the Ursel Clay Member.
caLcareous NaNNofossiLs General characteristics of the nannoflora The distribution of calcareous nannofossils in the Eocene of the Mont-des-Récollets is complex and directly related to the paleoenvironmental characteristics and burial history of the deposits. The absence of nannofossils in the Asse Clay Member and in the base of the Ursel Clay Member is probably due to local post-depositional (? Pliocene or Pleistocene) weathering, as these deposits are generally calcareous throughout the Belgian Basin. Their absence in the upper part of the Ursel Clay is an early diagenetic feature, as pyritic moulds of foraminifera and molluscs have been recognized in this interval in all sections studied in Belgium (King 1990). The upper part of the Vlierzele Member, the overlying Aalterbrugge Member, and Unit A1 of the Aalter Formation are also devoid of nannofossils (Fig. 10). This results from the estuarine/coastal character of these units, which are too coarse-grained or too marginal marine to yield interpretable nannofossil assemblages. Fortunately, the main part of the section is calcareous and contains poorly to fairly well-preserved assemblages. As a whole these are poor in numbers (less than 5 specimens/field of view in the lower half and between 5 and 10 in the upper half of the section) and in taxa (generally less than 25) (Fig. 10). They are marked by substantial selective dissolution, and, therefore, strong taphonomic bias. The highest species diversity (maximum of 35 taxa) is recorded in the upper part of the Brussel Sand Formation and in the Wemmel Sand Member. Small and middle-sized Noelaerhabdaceae (between 40% and 85%) and Coccolithus pelagicus (between 5% and 45%) dominate (Fig. 11B), as in most of the Paleogene nannofossil assemblages worldwide. The relative paucity in warm-water groups, such as Discoasteraceae (except for the lowermost two samples), Helicosphaeraceae and Sphenolithaceae, could be brought in relation to the middle to high-latitude position of the Mont-des-Récollets outcrop area during early and middle Eocene times (paleotemperature interpretation after Wei & Wise 1990). However, this seems to be an artifact, due to the erosional removal of the major part of Unit B3 at Mont-des-Récollets. This part of Unit B3, preserved in many outcrop sections around Brussels, includes a series of blooms of Pemma, Sphenolithus and Discoaster, all considered to be thermophilic taxa (Steurbaut, unpublished, detailed in a forthcoming study). The Brussel Sand Formation was governed by a shallow marine deposition regime, with a nearby coastline, as evidenced by the relatively high numbers of Zygrhablithus bijugatus, Micrantholithus and Pontosphaera, throughout the section, and of Lanternithus minutus, Blackites creber and Pemma spp. in certain intervals (Fig. 11B). Some particular levels (bases of B4 and the Lede Formation) are very rich in Braarudosphaera. Braarudosphaera bigelowii, which is known to occur in hyposaline turbulent turbid shallow water (Bukry 1974;Takayama 1972;Moshkovitch & Ehrlich 1982) and B. stylifera may reach up to 13% of the assemblage just above the basal sandstone of the Lede Formation.
One hundred and two different taxa have been identified at the Mont-des-Récollets, ten of which represent new species. Two of these belong to Luminocanthus n. gen., a new genus introduced herein. The stratigraphically important taxa are shown in Fig. 11A, those with quantitative importance in

Mont-des-Récollets
fig. 10. -Quantitative analysis and biozonation of the upper Ypresian and lower Lutetian calcareous nannofossil assemblages of the Mont-des-Récollets section (events 1 to 7 within NP 14 refer to the standard sequence of events, as identified in the present paper). Note that the occurrence of event 1 is slightly delayed at Mont-des-Récollets compared to the standard sequence of events. This seems to be due to the poor preservation of the assemblages in Unit A4 at this site.

Calcareous nannofossil distribution throughout the Mont-des-Récollets section (Figs 10; 11)
The assemblages from the Hyon Sand Formation are well diversified (around 30 taxa/sample) but very poor in number of specimens (1 to 3 per field of view at magnification 1000×) (Fig. 10). They are dominated by Coccolithus pelagicus (20-30% of the total number of specimens), Discoaster kuepperi (c. 20%), Toweius occultatus (c. 13%) and Pontosphaera pulchra (up to 11%), all four together making up about 65% of the assemblage (Fig. 11B). The co-occurrence of Discoaster lodoensis and Tribrachiatus orthostylus allows their attribution to nannofossil zone NP12, more specifically to its top part, according to the presence of Discoaster cruciformis and Nannoturba robusta (subzone VIIIb of Steurbaut 1998).
The Vlierzele Member is less rich in nannofossils. The upper 5 m are completely devoid; the lower 7 m are relatively poor (c. 20 taxa/sample, richest levels around 4 specimens per field of view). C. pelagicus (c. 40%) and Pontosphaera pulchra (between 15 to 27 %) dominate the assemblages. There is a substantial increase in small Noelaerhabdaceae up section (from c. 5% to 36%). As T. orthostylus is no longer encountered and Discoaster sublodoensis not yet present, the associations are attributed to zone NP13. According to outcrop and borehole data from N Belgium (Steurbaut 1990 and unpublished, see also Figs 12, 13) they seem to belong to the middle part of this zone (subzone NP13-b, as defined herein): co-occurrence of Nannoturba robusta, N. spinosa, N. jolotteana and Imperiaster sp. and only few Micrantholithus aequalis and M. inaequalis.
The overlying Aalterbrugge Member and the basal Unit A1 of the Aalter Formation do not yield nannofossils. They recur at the base of Unit A2, although in very low numbers. Assemblages throughout this unit remain poor (10 taxa/ sample, 1 specimen per field of view), except for these in the interval between the Venericor shell bed (Ve) and the turritelline coquina (Tu) (Fig. 7), that present a much higher species diversity (19 taxa/sample, 1 specimen per field of view). The assemblages are quite similar to these of the Vlierzele Sand Member, although richer in Noelaerhabdaceae (c. 50% of number of specimens), and also belong to zone NP13. The only substantial difference is that Nannotetrina cristata (LO within turritelline bed Tu) and Girgisia (formerly Toweius) gammation (LO in interval between Tu and Ve, see Fig. 7) have their first occurrences in Unit A2, while Imperiaster is no longer present (Fig. 11A). The co-occurrence of N. robusta, N. spinosa and N. jolotteana allows its correlation to subzone NP13-b.
The assemblages of unit A3 are quantitatively poor (often less than 1 specimen per field of view), but rather well diversified (c. 13 taxa/sample). They are attributable to the upper part of NP13 (subzone NP13-c, as defined herein) and marked by the disappearances of Nannoturba spinosa and N. jolotteana. The most diversified samples (16 taxa) occur between sandstone beds A and B (Fig. 7). Discoaster cruciformis, a typical late Ypresian form, disappears just below a conspicuous glauconite bed, a few meters above the base of A3 (110 m in Western quarry). It almost coincides with the lowest consistent occurrence of Reticulofenestra aff. umbilica (small forms: largest diameter c. 8 to 9 µm) and an increase of Chiasmolithus specimens. Toweius occultatus, another typical Ypresian form, disappears around the top of A3 (1 specimen recorded in the base of A4) (Fig. 11A). The assemblages in the upper part of Unit A3 (above bed B) are furthermore marked by the income of Zygrhablithus bijugatus crassus (111.70 m) and the presence of Ectalithus (formerly Coccolithus) crassus.
Unit A4 is slightly richer in nannofossils (between 15 and 25 taxa/sample, generally 3/maximum 5 specimens per field of view). Major changes are recorded within its lowermost metre, including the LO of Martiniaster cecellanoriae n. sp., the LO of Trochoaster nodosus n. sp., the LCO of Nannotetrina cristata and the HCO of Discoaster lodoensis at 114.85 m and the LOs of Discoaster praebifax and Pemma spp. slightly higher up (at 115.10 and 115.60 m repectively). Typical 5-rayed Discoaster sublodoensis specimens are consistently recorded from the middle of Unit A4 (116.45 m). The LOs of Blackites minusculus n. sp. and Sphenolithus recolletensis n. sp. are at the same level (116.45 m).
The assemblages of the lower part of the Brussel Formation (Unit B1) are similar to these of the top of the Aalter Formation, except for a somewhat lower species diversity (c. 17 taxa/ sample) and a higher proportion of Noelaerhabdaceae in B1. The same species content is recorded in Unit B2, which, however, shows a decrease in number of specimens and, although less severe, in species diversity (maximum 15 taxa/sample). The LO of Discoaster wemmelensis is just bove the base of Unit B1. It is the earliest record at the Mont-des-Récollets, but not in the Belgian Basin as this species is known to occur in low numbers in the top of the underlying Aalter Sand Formation (one specimen in the Oedelem BH and two in the Vlakte van de Raan BH) (Fig. 12). D. wemmelensis remains extremely rare up to the middle of B2. Higher up, from the upper part of Unit B2 onward, it is more consistently recorded and coincident with the LOs of Blackites praeinflatus n. sp. and Sphenolithus spiniger. The associations in Unit B3 considerably differ from the underlying ones through a sudden influx of Luminocanthus plenilutetiensis n. gen., n. sp., some last records, such as Girgisia gammation and Trochastrites pyramidalis n. sp., and a much higher species diversity (30 to 35 taxa/sample). Small Noelaerhabdaceae dominate the associations (80%). Substantial changes also occur at the base of Unit B4, marked by the lowest occurrences of Braarudosphaera stylifera, Martiniaster fragilis, Sphenolithus quadricornutus n. sp. and the LCO of Lanternithus minutus and Toweius brusselensis. This unit is also distinguished from the underlying units by the sharp rise in Braarudosphaeraceae (including Braarudosphaera, Micrantholithus and Pemma: from <1% to 17-22%), Lanternithus minutus (from 0% to 6-14%), Blackites creber (from 1% to 9-11%) and Zygrhablithus bijugatus (from 1% to 13-16%).
The assemblages above the basal limestone of the Lede Sand Formation are quite similar to these of Unit B4, in frequency (7 specimens/field of view) as well as in composition, as they are made up of the same dominant groups: Braarudosphaeraceae (among which B. stylifera) = 19%, Z. bijugatus = 12%, B. creber = 11% and L. minutus = 8%. The assemblage of the non-lithified lowermost 5 to 10 cm of the Lede Formation GEODIVERSITAS • 2021 • 43 (11) (below the basal limestone) is richer in small Noelaerhabdaceae, but less rich in Braarudosphaeraceae (6%) and L. minutus (1%). Among the main differences with Unit B4 are the first, but rare, record of Blackites gladius and the disappearance of D. sublodoensis at the base of the Lede Sand Formation. The assemblages remain fairly unchanged throughout the Lede Formation, except for the disappearance of Toweius brusselensis just above the basal sandstone and the frequency in Braarudosphaeraceae, which systematically decreases up section (from 19% at 123.50 m to 10%, 9% and 7 % at respectively 124.10, 128.55 and 129.55 m).
No nannofossils were encountered in the coarse-grained glauconitic base of the Wemmel Sand Member ('Bande noire' sensu Lyell 1852). The overlying beds (still part of the base of the member) yield fairly rich assemblages (between 30 and 35 taxa/sample; between 2 and 7 specimens/field of view), belonging to the middle part of NP 15 (presence of Blackites gladius). They differ from these of the underlying Lede Formation (lower NP 15) by the decrease in Braarudosphaera stylifera (a few specimens are present in the base of the member), the LO of Sphenolithus furcatolithoides and Naninfula deflandrei, and the LCO of Pontosphaera wechesensis.
At the Mont-des-Récollets the assemblages of the overlying Asse Clay Member are strongly dissolved, and cannot be dated satisfactorily. This is also true for this interval in the Cassel borehole (Lezaud 1972). Sullivania gigas (Coccolithus gigas sensu Lezaud 1972) has not been identified in the present study. Its presence in the Lede Formation and the base of the Maldegem Formation in the Cassel borehole, as claimed by Lezaud (1972), is doubted and believed to be a misinterpretation.

Identification of Martini's NP-Zones and subdivision of Zone NP13
Martini's (1971) standard calcareous nannofossil zones NP12, NP13 and NP14 are easy to recognize in the sections of the southern North Sea Basin, including Mont-des-Récollets ( Fig. 10), because of the frequency of Discoaster lodoensis, Tribrachiatus orthostylus and Discoaster sublodoensis, species which are used to define the lower boundaries of these zones. However, although the presence of Zone NP14 is easy to establish, the identification of its base is less straightforward in the southern North Sea Basin. This is due to the scarcity of typical five-rayed specimens of Discoaster sublodoensis at the start of its range. Fortunately, the LO of the latter coincides with the LO of many other species and with the HO of Toweius occultatus. This major bio-event is defined herein as the Base Lutetian calcareous nannofossil turnover (abbreviated as BALCAT), marked by fundamental changes in the shallow-water nannofossil assemblage (see next chapter).
The Mont-des-Récollets data allow subdivision of Zone NP13 into three subzones (Fig. 10). The LOs of Nannoturba jolotteana and Nannoturba spinosa, which are coeval and clearly postdate the HO of T. orthostylus (representing the base of Zone NP13 or NP13-a), are used to define the base of a second subzone NP13-b. Their HOs, which are also coeval, allow to define the base of a third subzone NP13-c. The events used to subdivide NP13 have been identified throughout the Belgian Basin (see biostratigraphic interpretation of the assemblages and chapter 'Discussion' below) and may be excellent candidates for enhancing interbasinal correlations within the Northwest Atlantic realm. N. spinosa, initially described by Müller (1979)  The lower boundary definition of NP15, is also difficult to apply in the southern North Sea Basin, because of the scarcity of the marker species Nannotetrina alata. Up to now, only two specimens have been recored in Belgium, one from the Zaventem-Brussel Airport section (Steurbaut in Herman et al. 2001) and one from the Nederokkerzeel outcrop (Steurbaut pers. comm.). Identification of zone NP 15 is based on the LO of Nannotetrina quadrata (Steurbaut in Smith et al. 2004), or on the LO of Blackites gladius (Steurbaut 1990), which according to Martini (1971) and Aubry (1983: 105, based on data from the North Atlantic Ocean) are coincident or very close to the LO of N. alata. The additional Nannotetrina taxa (N. fulgens and N. pappii) occur slightly later, although still in NP15 (in Belgium, in the upper part of the Wemmel Sand Member and the base of the Asse Clay Member respectively; Steurbaut et al. 2015).
A major turnover in shallow-water calcareous nannoflora at the base of NP14 The calcareous nannofossil investigation of the Mont-des-Récollets section reveals a fundamental change in assemblage composition at the base of Unit A4 of the Aalter Sand Formation. This change is marked, among many other things (see below), by the transition from Discoaster lodoensis to Discoaster sublodoensis. The latter is abrupt, apparently taking place during the hiatus between Unit A3 and Unit A4. It includes several stepwise transformations. Typical multirayed (5 to 8) specimens of Discoaster lodoensis are characterized by curved rays with the conspicuous peripheral ridge, running along the less curved edge of the ray towards the centre. At first, specimens with almost straight rays are popping up, while keeping all other characteristics of D. lodoensis (this form is defined herein as the 'late form' of D. lodoensis). Subsequently, these D. lodoensis specimens with straight rays transform into six-rayed discoasters without peripheral ridges (known as Discoaster strictus and allies) and finally into a five-rayed form. The latter has much shorter rays and is marked by a small low stem and low crests along the margin of the rays, ending at the interstices between the rays. This five-rayed form is considered herein to represent D. sublodoensis (Fig. 18I-M), and its LO is used to define the base of NP14, as already suggested by Agnini et al. (2006Agnini et al. ( , 2014. The assemblages at the base of Unit A4 are fairly rich in discoasters, essentially consisting of the 'late form' of Steurbaut E. & Nolf D.    D. lodoensis (form with practically straight rays). The typical form of D. lodoensis with curved rays as well as D. strictus are also present, although very rarely and are co-existing with very rare five-rayed specimens of D. sublodoensis. The extreme rarity of D. sublodoensis at the start of its range may hamper the identification of the base of Zone NP14. Fortunately, the LO of D. sublodoensis is linked to a major change in calcareous nannofossil composition, essentially due to diversification in shallow-water groups. This is clearly expressed in the well-preserved nannofossil assemblages of the Vlakte van de Raan BH in Belgium and in the 'Chaumont-en-Vexin sands' (term informally introduced herein) in the Paris Basin, but is less obvious in the poorly preserved assemblages of the Mont-des-Récollets. It consists of a series of lowest occurrences among which are these of Blackites minusculus n. sp., Blackites praeinflatus n. sp., Lanternithus minutus, Luminocanthus eolutetiensis n. gen., n. sp., Martiniaster cecellanoriae n. sp., Trochoaster nodosus n. sp. and a major radiation in Trochastrites. Due to its protective cap rock, preventing substantial post-Miocene erosion, and its key position in the tectonically relatively quiescent southwest edge of the Belgian Basin (Fig. 1), the Mont-des-Récollets outcrop has preserved a very complete upper Ypresian and Lutetian succession. Its particular geological situation allows delineating more accurate spatial distribution patterns of the Ypresian and Lutetian lithostratigraphic units, which were nearly all defined in areas north and east of the Mont-des-Récollets (Steurbaut et al. 2015(Steurbaut et al. , 2016a. The identification of the Mont-Panisel Sand Member on the southern flank of the Mont-des-Récollets is not really a surprise, as this unit is known from many outcrops and boreholes along the southern edge of the Belgian Basin (e.g. Mont-Panisel at Mons, Steurbaut & King 1994; Mont-Saint-Aubert, North of Tournai, Steurbaut unpublished information). However, this is the first undoubted record of the Mont-Panisel Sand Member in France. The detection of high frequencies of Discoaster kuepperi (over 20%) (Fig. 11B) is in line with earlier records at Zemst (central Belgium), where these peak abundances were linked to the development of warm sea surface waters during the Early Eocene Climatic Optimum (EECO) (Steurbaut et al. 2015). Identical lithologies, also belonging to upper NP12 (co-occurrence of Tribrachiatus orthostylus and Discoaster cruciformis) have been described from the nearby (1.5 km) Cassel borehole (interval c. 83 m to c. 77 m), but were erroneously attributed to the 'Argile de Roubaix ' (Blondeau et al. 1972).
In the Cassel borehole the Mont-Panisel Member is overlain by a 3 m thick stiff clay of which the base contains Palaeonummulites aquitanicus (N. planulatus auctoris) and the uppermost 1 m is slightly calcareous, indicating NP13 (attributed to the 'faciès d'Aeltre azoïque' by Blondeau et al. 1972: tab. 1). Its attribution to the Merelbeke Clay Member is supported by its typical lithology and stratigraphic position, implying that this Merelbeke Clay record is the first of its kind in France. A few years earlier, in 1968, a similar 2.5 m thick clay unit, termed P1m, has been described from the Rodeberg borehole (municipality of Westouter, W Belgium, see Appendix 1), 18 km to the east of the Mont-des-Récollets (57.7-60.0 m; Gulinck unpublished information). Although it has surprisingly complete sections, the underlying Kwatrecht Member could not be unequivocally identified in any of the discussed outcrops or boreholes of the Mont-des-Récollets area (possibly recorded at the Rodeberg between 60 and c. 62.5 m depth). Nevertheless, these new records and interpretations indicate that the Merelbeke Clay Member covers a much vaster area of the Belgian Basin than previously thought, probably as far as the Artois Anticline in northern France, but that in its eastern direction it did not overstep the Brussels-Turnhout area (recorded at Kester and Merksplas: Steurbaut et al. 2016a). This is also true for the Pittem Clay Member, which is shown to be present in the Cassel borehole (72.30-71.40 m) and the Rodeberg borehole (57.7-48.5 m). A 20 cm thick shelly glauconite-rich sandstone with small pebbles, very similar to the Hooglede Bed in the Egem quarry, about 56 km northeastward (Steurbaut 2006), has been recorded at the base of the Pittem Member in the Rodeberg borehole. About 12 m of the overlying Vlierzele Sand Member is exposed in the 'Western Quarry' of the Mont-des-Récollets, attributed to Units Vli b (essentially alternation of sand and clay beds) and Vli c (essentially bioturbated or low angle laminated glauconitic fine to very fine sand). The increase in quartz grains and glauconite towards the base of the section might suggest that the top of Unit Vli a (loose bioturbated fine sand) has been penetrated or is very close. These units are part of the threefold subdivision of the Vlierzele Sand Member, as observed in the Oedelem borehole (7 km east of Brugge), where this unit is remarkably complete (c. 13 m thick, Fig. 12 at the Mont-des-Récollets. It probably includes the complete suite of members (only the presence of Kwatrecht Member is questionable), with lithologies very comparable to those in their respective type areas in central Flanders (essentially the province of East Flanders). The combined lithofacies and calcareous nannofossil investigation given herein has revealed that the Aalter Sand Formation is very expanded (18.25 m) at the Mont-des-Récollets (Fig. 14A). Its lithofacies succession shows strong affinities with the stratotype section at Aalter (12.35 m thick; Steurbaut & Nolf 1989) and with the Hijfte borehole section (c. 10 m thick, Fig. 13), although has its own particularities. The Mont-des-Récollets Units A1 and A2 contain Venericor planicosta beds, one bed in each unit, with turritelline coquinas in between. This contrasts with the Aalter stratotype where the turritelline coquinas overlie the interval with V. planicosta (Steurbaut & Nolf 1989). In the Oedelem borehole this lower part of the Aalter Formation is much finer and clayier (clayey very fine sand), representing the Beernem Sand Member. The correlation between the top of Unit A2 and the top of the Beernem Sand Member at Oedelem is established on the basis of the synchronous disappearances of Nannoturba jolotteana and N. spinosa (Fig. 12).
The base of Unit A4 corresponds to a sequence boundary, which can be traced throughout the Belgian Basin (Aalter stratotype, Oedelem borehole, SEWB borehole off Blankenberge, etc.; base of dark green coloured interval in Figures 12  and 13 and Appendix 1 for locations). At Aalter and Montdes-Récollets this contact could not be properly investigated because of the rudimentary drilling technique (shallow auger boreholes), but in the SEWB offshore borehole it is marked by an undulating omission surface at 19.17 m depth (Jacobs & Sevens 1993: fig. 10). In each of the investigated sites this sequence consists of a thin glauconitic fining upward fine sand to clayey fine sand, overlain by a much thicker coarsening upward sandy succession (Unit A4 at Mont-des-Récollets; bed 12 in the Aalter stratotype section: Steurbaut & Nolf 1989;units 19 and 20 of SEWB hole: Jacobs & Sevens 1993). The base of Unit A4 is marked by a major change in calcareous nannofossil composition, including many lowest occurrences (among which the LO of D. sublodoensis) and the HO of Toweius occultatus (see previous chapter). This turnover at the base of NP14 is well expressed in the well-preserved nannofossil assemblages of the Vlakte van de Raan BH in Belgium's offshore zone and in the 'Chaumont-en-Vexin sands' in the Paris Basin, but is less obvious in the poorly preserved assemblages of the Mont-des-Récollets. However, the highest consistent occurrence (HCO) of Discoaster lodoensis and the lowest consistent occurrence ( depth in the Leuven-Artois borehole, represents a sequence boundary, and, more precisely, the same as the one recorded at the base of Unit A4 at Mont-des-Récollets and at the base of bed 12 in the Aalter stratotype (Steurbaut & Nolf 1989). This means that, at least in the Leuven area, the start of the major scouring event, producing the Brussels megachannel (Houthuys 1990;Damblon & Steurbaut 2000), dates from the start of Biochron NP14, around 49.11 Ma (according to Payros et al. 2015). This megachannel is in fact a deeply incised 25 km wide estuary, marked by several tens of meter deep ebb-and flood-dominated channels (Fig. 14B). It also indicates that the lower part of this channel fill is coeval with the upper part of the Aalter Sand Formation. The contact of the Aalter Sand Formation with the overlying Brussel Sand Formation has not yet been studied in much detail, as both units are only rarely recorded in superposition (e.g. in the Flemish Hills and in some offshore boreholes) or because of the many coring difficulties related to the sandstone levels in the latter (e.g. Kallo and Mol boreholes). The typical whithish medium-grained sand with echinoid (among which Maretia omaliusi) and bryozoal fragments, occurring between 117.35 m and 118.95 m at the Mont-des-Récollets, are included in the Brussel Sand Formation (Unit B1). However, there is no omission surface at their base, as already observed by Leriche (1912: 713); there is only a sudden increase in grain size, and the nannofossil assemblage is very similar to that at the top of the underlying Aalter Formation. This contrasts with the boundary between Units B1 and B2, which is marked by an omission surface, directly overlain by very coarse quartz sand. At the Mont-des-Récollets the LO of Discoaster wemmelensis is recorded at the base of B1, whereas the LOs of Blackites praeinflatus n. sp. and Sphenolithus spiniger and the HO of S. recolletensis n. sp. are about 2 m higher up, in the upper part of the bioturbated sands of Unit B2b. There is an interburrowed junction and a sharp decrease in grain size at the base of Unit B3, which coincides with the influx of Luminocanthus plenilutetiensis n. gen., n. sp. Comparison with the nannofossil assemblages from the offshore Vlakte van de Raan borehole section and different outcrop sections in the Brussels area (e.g. at Neerijse, Sint-Stevens-Woluwe, Vossem, etc.; see Figs 12, 13 and Appendix 1 for locations) indicates that D. wemmelensis, L. plenilutetiensis n. gen., n. sp. and B. praeinflatus n. sp. show somewhat delayed first records at the Mont-des-Récollets, probably due to the low quality and low diversity of the assemblages in Units A4 and B1 (Fig. 10). It also suggests that initially Unit B3 was probably much thicker at the Mont-des-Récollets and that this unit was completely removed during the following transgression, except for a thin 40 cm thick sedimentary succession, left as relict (Fig. 9). This is evidenced by the absence of the Sphenolithus, Pemma and Discoaster blooms at Mont-des-Récollets, which in many outcrop sections (Diegem, Vossem, Zaventem,…), start a little above the base of Unit B3 (Steurbaut unpublished, detailed in forthcoming study). Unit B4, the base of which is sharp and believed to be erosional, is characterized by the frequent occurrence of Nummulites laevigatus (c. 5 specimens/ kg sediment). It is one of the very few in situ occurrences of this taxon in the Belgian Basin (see next chapter). The nannofossil assemblages are also quite distinct, marked by rather high values of Braarudosphaera bigelowii, B. stylifera, Pemma spp., Lanternithus minutus and Blackites creber (Fig. 11B). This points to substantial paleoenvironmental changes within the Belgian Basin, with a tendency to develop coastal hyposaline conditions along its southern border. The base of Unit B4 is also a major biostratigraphic boundary, marked by the LOs of B. stylifera, Martiniaster fragilis and Sphenolithus quadricornutus n. sp. and by the LCO of L. minutus.
At the Mont-des-Récollets the texture, composition (presence of several sandstone banks) and thickness of the Lede Sand Formation is very similar to that of the stratotype section at Balegem (Jacobs & Sevens 1994). At both sites the basal surface is erosive and overlain by a basal gravel including reworked shell fragments, pebbles and shark teeth. However, there is a fundamental difference in the reworked shell fragments, which essentially consist of loose shells and loose or consolidated aggregates of Nummulites laevigatus at the Mont-des-Récollets (see next chapter) and of Venericor planicosta and very few N. laevigatus at Balegem (Fobe 1986: 15). At both sites, the nannofossil assemblage of the basal Lede Sand Formation is quite similar to that of B4. Other similarities between both sites are the rare records of absence of Campanile giganteum in the Lede Sand Formation in Belgium (only 2 specimens recorded, see below), while it is consistenly recorded in this unit at the Mont-des-Récollets, although restricted to a single bed (bed 4 on Fig. 9).
At the Mont-des-Récollets the Wemmel Sand Member is surprisingly similar to what has been observed in the centre of Gent (Blandijnberg or 'Colline de Saint-Pierre': Delvaux 1886; Nolf 1974), including a 10 cm basal glauconite bed, a 20 cm thick clay bed, a 10 cm thick accumulation of Lentipecten, overlain by clayey fine sand. At the Mont-des-Récollets the basal clays and Lentipecten bed contain fairly rich nannofossil assemblages, marked by the LO of Sphenolithus furcatolithoides and the HO (a few specimens) of Braarudosphaera stylifera. Blackites gladius is present, but rarely recorded, while Nannotetrina taxa are absent. Identical assemblages have been observed at the base of the Wemmel Sand Member in the Zemst borehole (Steurbaut et al. 2015;Lin et al. 2017 fig. 4) were among the first to mention their presence at Mont Cassel and Mont-des-Récollets. Van den Broeck (1902) discussed the reworking of N. laevigatus shells at the base of the Lede Sand Formation (= base Laekanian sensu Van den Broeck). He stressed on the presence of two types of N. laevigatus, silicified forms without glauconite infill co-occurring with non-silicified, glauconitecontaining forms, which seems to point to different source areas. Leriche (1906: 395;1912: 711) suggested that probably all of the N. laevigatus records of Belgium and northern France were due to reworking, except for these at Cassel and Fayat (between Charleroi and Namur), which were believed to be in situ, as already suggested by Rutot (1887). In his subsequent review of the spatial distribution of N. laevigatus in the Belgian Basin Leriche (1923) confirmed that the occurrences in the Flemish Hills (Mont Cassel, Mont-des-Récollets, Mont des Cats and Mont Aigu, see Fig. 1) and in the zone between Leuven-Charleroi-Namur represented in situ records. He expressed serious doubt about other previous records, such as these at Aalter, Gobertange, Westerlo, etc., and did not mention any record from the Brussels area (Leriche 1923: 94). Blondeau (1966) listed several additional sites, especially around Brussels (Forest, Diegem) and presented some morphometric data on specimens from Jodoigne and Isnes, located respectively 28 km north and 10 km northwest of Namur. The in situ records in the Woensdrecht borehole (southern Netherlands) (Van Waterschoot van der Gracht et al. 1913) and in Zeebrugge borehole 19.6 (Depret & Willems 1983) provide the necessary evidence to refute Blondeau's hypothesis (Blondeau et al. 1972) that the Flemish Hills and the Brussels area define the northern limit of the distribution of N. laevigatus in the Belgian Basin.
Our personal observations at Gobertange (Damblon & Steurbaut 2000) and Zaventem (Herman et al. 2001), as well as those from other geologists at Diegem (Fobe 1986; Hooyberghs 1986), Nederokkerzeel (Fobe 1986), Neerijse (Fobe 1986;Hooyberghs 1992) and Sint-Stevens-Woluwe (Hooyberghs 1992), which did not lead to in-situ finds of N. laevigatus, indicate that this species is rarely recorded in situ in the Brussels area. However, there is sufficient evidence to assume that a thin N. laevigatus bearing sand and/or sandstone layer(s) with abundant large microspheric forms of N. laevigatus, covered the greater part of the Belgian Basin (equivalent of the 'assise de la pierre à liards' in the Paris Basin), as already postulated by Gosselet (1883), Leriche (1906: pl. VII) and Dartevelle (1934). We also agree with their conclusion that during the deposition of the N. laevigatus beds the Belgian Basin was directly connected with the Paris Basin. The N. laevigatus beds were almost completely dismantled in central Belgium, due to substantial tectonic uplift of the Brabant Massif, whereas in the tectonically relatively quiescent zones (e.g. the Flemish Hills) parts of these remained preserved. Moreover, in the north of the province of Namur (Nalinnes, Isnes, etc.), erosion of the N. laevigatus bearing beds was least pronounced or was even neutralized by local subsidence, as shown by their strong increase in thickness, up to more than 2 m at Nalinnes (Briart 1890: 263) and more than 3 m at Saint-Denis-lez-Namur (Halet 1937), Noville-sur-Mehaigne (Leriche 1943b) and Isnes (Hooyberghs 1986: fig. 11).

Nummulites laevigatus-bearing sandstone blocks at the base of the Lede Formation
The abundance of N. laevigatus-bearing sandstone blocks in the base of the Lede Sand Formation (also known as base Laekenian in older literature) has systematically been reported from the Mont-des-Récollets for almost two centuries (Rutot 1882;Leriche 1921;Depret & Willems 1983;and Fobe 1986, for overviews). Some of these blocks, such as the one displayed herein (Fig. 15), are composed of different lithofacies types, witnessing multiple lithification phases, and as a whole, revealing a complex depositional and tectonic history of this part of the Belgian Basin.
The core of the sandstone block studied herein consists of an amalgamation of abundant, almost unidirectional oriented large-sized N. laevigatus shells (Fig. 16), a mixture of essentially fine-grained and some coarse-grained glauconite (the latter in burrows ?), some larger quartz grains (1 mm), within a finegrained matrix, all solidly cemented (probably ferro-calcitic cement) (phase 1). The presence of large grains may be due to bioturbation (burrowing molluscs; B on Fig. 15). This core is irregularly surrounded by a thin greenish coloured coarse sandstone, less well consolidated and containing a mixture of Nummulites laevigatus, Palaeonummulites variolarius, some small molluscs, echinoid debris and Ditrupa, and above all marked by huge (up to 2.5 mm) quartz grains and coarsegrained glauconite (phase 2). The upper zone consists of a rusty stained, finer grained sandstone, presenting an irregular basal surface, and containing abundant fine-grained glauco-GEODIVERSITAS • 2021 • 43 (11) nite, large molluscs, P. variolarius, Ditrupa and bryozoa, but only very few N. laevigatus (phase 3).
The N. laevigatus-rich core appears to have been formed during a first lithification phase of initially relatively finegrained nummulitic shallow marine sand of the Brussel Sand Formation. This lithification took place under continental conditions, probably shortly after the sea retreated. Remains of this continental phase have been reported from the base of the Lede Formation at the Mont-des-Récollets by Leriche (1921), who mentioned the presence of reworked lignitic pebbles and crocodile remains. During the next sea-level rise the continental cover as well as the underlying N. laevigatusbearing sandstones and associated badly sorted sand with large quartz grains were eroded. The most solid elements (rolled sandstone blocks) were abraded and colonized by lithofageous bivalves. These blocks resedimented together with reworked loose N. laevigatus and many in situ fossils, such as P. variolarius, several mollusc species and Ditrupa tubes. Initially the winnowing of sediment by coastal currents and storm events was probably important enough to restrict sediment accumulation, explaining its coarse-grained texture (phase 2). In a later stage, when sea-level progressively rose, the winnowing decreased, which led to the deposition of a somewhat finer sediment (phase 3), rich in molluscs, Ditrupa, bryozoa and P. variolarius. Coastal conditions remained, with very reduced sediment supply, initiating compaction and, when the sea temporarely retreated, the start of calcite cementation of the sea bottom, including the earlier lithified N. laevigatusbearing boulders. Because marine and non-marine conditions rapidly shifted, the coarse-grained and finer grained sandstone components of phases 2 and 3, are poorly cemented, much less than the cementation in the core of the block (phase 1). Subsequently, when sea-level further rose, the effect of the coastal and storm currents reduced, initiating the progressive accumulation of fine shallow marine sand of the Lede Sand Formation. This was probably caused by a major storm event, intensifying the normal coastal currents. The Compiègne High and the Pays de Bray Isle may have protected the more southwestern areas of the Paris Basin (Yvelinnes and Paris areas) from these storms, or at least may have strongly diminished their effect, as shown by the scattered occurrences of C. giganteum in these areas.

Campanile giganteum in the Lede Sand Fm and its paleoenvironmental consequences
Records outside the Paris Basin are very rare and need to be interpreted with caution. Jung (1987) mentioned three incomplete specimens from the middle Eocene of the Lesser Antilles (Caribbean Islands), which he referred to Campanile cf. giganteum. Much closer, in the North Sea Basin, these giant gastropods have systematically been reported from the Bracklesham Beds of the Hampshire Basin (Fisher 1862; Wrigley 1940; amongst many others), although their exact position has been uncertain for a long time. Fisher's Bed 12 or the Campanile Bed was relocated at Bracklesham Bay by C. King (in Curry et al. 1977), at about 1.75 m above the base of the Selsey Division (unit S2i). About 20 years later King (1996) rediscussed its position and equated Fisher's Bed 12 with unit S4i. Almost simultaneously, Tracey et al. (1996) tabulated the stratigraphic distribution and relative abundance of the molluscs within this Selsey Formation. C. giganteum was shown to be restricted to units S2 to S4. In southern England this species is known to co-exist with Nummulites variolarius (King 1996: text- fig. 1). Its range clearly postdates that of N. laevigatus, very similar to what has been encountered in the Paris Basin (Gély 1996: fig. 2) and at the Mont-des-Récollets (Figs 9; 12; 13).
Occasional records (moulds) of C. giganteum have regularly been reported from the Eocene of the Cassel area, already since the earliest stratigraphic observations (Elie de Beaumont 1833; d'Archiac 1839; Lyell 1852; etc.). According to Ortlieb & Chellonneix (1870: fig. 9) they are restricted to a single limestone bank in the lower part of the 'Assise Laëkenienne' (bed 17) at the Mont-des-Récollets. This marker bed exactly corresponds to our limestone bed 4 in the lower third of the Lede Sand Formation (Fig. 7). The correlation between this limestone bed and the C. giganteum-bearing beds in the Paris Basin was already established by Gosselet (1896: 164). The presence of C. giganteum in Belgium has been confirmed, although from two localities only and without any detail. Burtin (1784: 106, pl. 16G) and Galeotti (1837: 36 and 59) mentioned a find from Affligem (18 km west of Brussels), but it is unclear if both cited the same specimen or specimens, whereas Wesselingh et al. (2013: fig. 1H) reported 1 specimen from the Lede Formation of the Balegem area (collector unknown). A third internal mould has been dredged from the Flemish Banks, offshore the northern Belgian Coast (Wesselingh et al. 2013: 160). According to the available data it is logically to assume that all the Belgian finds come from the Lede Sand Formation.
From the foregoing it is clear that Campanile giganteum is present in the lower part of the Lede Sand Formation, but with strongly decreasing numbers towards the north. This suggests a direct connection between the Paris Basin and the Belgian Basin during early Biochron NP15 (middle Lutetian) and the development of similar paleoenvironmental conditions in Steurbaut E. & Nolf D. these areas. The sedimentology of the sandy limestone banks in the Lede Formation indicates reduced sedimentation rates, storm-induced winnowing and hardground formation under a relatively thin water cover (<20 m, probably much less) (Fobe 1986: 177-180;Jacobs & Sevens 1994). This type of paleoenvironment is comparable to the current habitat of the only survivor of the family Campanilidae, C. symbolicum Iredale, 1917 from southwestern Australia. According to Houbrick (1984) large populations of this survivor taxon may be found on sandy patches between rocks in depths of one to four meters.
The spatial distribution of C. giganteum suggests that the shallow marine conditions, which led to the development of the Lede Sand Formation, persisted up to central Flanders, as far northward as the Aalter (outcrop)-Assenede (borehole 25E-123)-Kallo (borehole 27E-148)-Woensdrecht (ROD borehole 17, southern Netherlands) area ( Fig. 14C and Appendix 1 for details). Further northwestward, the Lede Sand Formation is missing (e.g. in the onshore Oedelem borehole and the offshore Vlakte van de Raan borehole in Belgium (Fig. 14C) as well as in the E-55 Rotterdam borehole in the Netherlands, Steurbaut unpublished information), most probably because of non-deposition, although subsequent erosion cannot be excluded either. The Aalter-Assenede area is likely to represent the most northern distribution limit of C. giganteum in the North Sea Basin. Its presence in dredgings from the offshore 'Flemish Banks', 50 km north, is believed to be due to northwest transport of reworked material from southern source areas via Quaternary river systems.
eNhaNciNg the upper ypresiaN-LutetiaN caLcareous NaNNofossiL record iN the southerN North sea basiN The calcareous nannofossil record of the Mont-des-Récollets outcrop allows fine tuning of the distribution patterns of many important index taxa, resulting in the subdivision of the standard NP13 and NP14 Zones of Martini (1971). This new succession of bio-events has been identified in many borehole and outcrop sections of the Belgian Basin. It has been recorded in the Bracklesham Group in southern England and, in part, in the lower Lutetian deposits of the Paris Basin, including the fine-grained 'Chaumont-en-Vexin sands' (term informally introduced herein) and in the overlying coarser grained 'Glauconie Grossière s.s.'. Combined with the Zemst borehole data, which led to the subdivision of Zone NP15 (Steurbaut et al. 2015), it is considered to be a standard for upper Ypresian-Lutetian calcareous nannofossil biostratigraphy of the southern North Sea Basin. Among the most relevant conclusions are: 1) Steurbaut's (1998) nannofossil subzone VIIIb, corresponding to uppermost NP12, has been identified at the Montdes-Récollets (Fig. 10), as well as in many other outcrop and borehole sections in the North Sea Basin (Steurbaut 2011) and Kazakhstan (King et al. 2013). The LO of Nannoturba robusta and the LO's of Ectalithus (previously Coccolithus) crassus and Pentaster lisbonensis, located at the base and in the upper part of the middle Ypresian subzone VIIIb respectively (e.g. at Zemst, Steurbaut et al. 2015 and Kerksken, Steurbaut unpublished information) are shown to be important datum planes in mid-latitudes of western Eurasia.
2) The acme of Discoaster kuepperi (around 20%), known from the Mont-Panisel Member (Hyon Sand Formation) in the Zemst borehole and associated with high sea surface water temperatures during the EECO (Steurbaut et al. 2015), is recorded in the same unit at the Mont-des-Récollets (Fig. 11B).
3) In the Cassel borehole the Merelbeke Clay Member is slightly calcareous at its top, indicating the lower part of NP13 (Lezaud 1972). This is the first calcareous nannofossil record of the Merelbeke Clay Member, which is secondarily decalcified elsewhere in the Belgian Basin (e.g. mollusc moulds at Roborst, Steurbaut unpublished information). It is in line with its generally accepted, but only indirectly established, biostratigraphic position (above the Kwatrecht Member dated as lower NP13) (Steurbaut et al. 2015).
4) The Mont-des-Récollets data allow subdivision of Zone NP13 into three subzones (Fig. 10), all recorded throughout the Belgian Basin. The coincident LOs of Nannoturba jolotteana and N.spinosa, defining the base of subzone NP13-b, have been identified at the base of the Pittem Clay in many Belgian boreholes (e.g. the Oedelem, Knokke and Kallo boreholes; see Figs 12, 13), although they were not mentioned in Lezaud's work on the Cassel borehole. Their coincident HOs are recorded at the top of Unit A2 in the Mont-des-Récollets section, as well as at the top of the Beernem Sand Member in the Oedelem borehole, indicating that these tops are coeval (base of subzone 13-c). Chiphragmalithus vandenberghei (see Steurbaut 2011) appears to be restricted to subzone NP13-b. Discoaster praebifax, Nannoturba joceliniae n. sp. and Pemma spp. first appear in subzone NP13-c, just above the base and in the middle (the last two) respectively.
5) The LO of typical five-rayed Discoaster sublodoensis, which is used to define the base of NP14, has been identified at the base of Unit A4 in the Vlakte van de Raan borehole and at the base of the 'Chaumont-en-Vexin sands' in the Paris Basin. Its LO is slightly delayed in the more poorly preserved assemblages of the Belgian Basin, such as at the Mont-des-Récollets, although still within the lowerpart of Unit A4. Fortunately, this bio-event is part of a major turnover in calcareous nannofossil assemblages, which is recorded throughout the Belgian Basin, including the Mont-des-Récollets, and in the Paris Basin. It involves the identification of many lowest occurrences (e.g. these of Blackites minusculus n. sp., Blackites praeinflatus n. sp., Lanternithus minutus, Martiniaster cecellanoriae n. sp., Luminocanthus eolutetiensis n. gen., n. sp., Trochoaster nodosus n. sp.), a major radiation of Trochastrites (the LOs of T. bramlettei, T. pyramidalis n. sp. and 3 other undescribed new species) and the HO of Toweius occultatus (a few isolated specimens of T. occultatus may linger on in the base of A4). The HCO of D. lodoensis and the LCO of D. sublodoensis, which occur higher up in Unit A4, are also basin wide phenomena.
7) The LO's of Nannotetrina alata (Zaventem section, Herman et al. 2001) and Nannotetrina quadrata (Oosterzele section, Smith et al. 2004) have been used to define the base of NP15 in the Belgian Basin, as Blackites gladius is often rarely represented in the earliest part of its range (e.g. at Mont-des-Récollets, Balegem and Oosterzele). At the Mont-des-Récollets, where Nannotetrina is missing in the Lede Sand Formation, the LO of B. gladius is used to define the base of NP15. The HOs of Discoaster sublodoensis and, to a lesser degree, Toweius brusselensis (still present in the basal 20 cm of the Lede Formation, but not higher up) may be considered as reliable substitutes for the lower boundary of NP15. 8) Sphenolithus perpendicularis, initially described from the middle Lutetian in offshore West Australia (throughout CP13a, lower part of NP15), has been identified in the middle of the Lede Sand Formation at Balegem. This is the first record of this species in the North Sea Basin. and 9) The subdivision of Zone NP15, as established in the Zemst borehole (Steurbaut et al. 2015;Lin et al. 2017) appears to be applicable to the Mont-des-Récollets, at least for the studied interval (up to mid-NP15). The clay bed and overlying Lentipecten bed, composing the base of the Wemmel Sand Member at the Mont-des-Récollets, are marked by the co-occurrence of Sphenolithus furcatolithoides and Braarudosphaera stylifera, allowing correlation with the top of subunit W1 of the Wemmel Sand Member at Zemst (Steurbaut et al. 2015: 152).

the ypresiaN-LutetiaN bouNdary iN the beLgiaN basiN aNd its iMpact oN gLobaL chroNostratigraphic cLassificatioN
A close inspection of the relevant outcrop and borehole data of Belgium (Steurbaut 2006(Steurbaut , 2011Steurbaut et al. 2015;Figs 12, 13) reveals that the Ypresian-Lutetian transition is very incomplete in large parts of the Belgian Basin. Complete or almost complete Ypresian-Lutetian transition sections may have been preserved north of the Antwerp area and in the southern part of the province of Zuid-Brabant in the Netherlands (Woensdrecht borehole;Halet 1913;Van Waterschoot et al. 1913) (Fig. 1). However, this is difficult to evaluate or to confirm because of substantial core loss and related displacements of core material at the contact of hard banks and sandy sediments, which are quite common in that interval. In the northwest edge of the Belgian mainland (Zeebrugge borehole 19, Depret & Willems 1983 and Knokke borehole, King 1990) and its prolongation offshore ('Vlakte van de Raan' borehole, Fig. 12) the Brussel Sand Formation is represented by very fine glauconitic sand, in the offshore borehole erroneously attributed to the Wemmel Sands (Jacobs & Sevens 1993;De Batist & Henriet 1995), while the Lede Sand Formation is totally missing. At the Mont-des-Récollets both formations are very well developed and the underlying Aalter Sand Formation is among the most complete ever recorded, justifying the peculiar status of the Mont-des-Récollets section as 'the' reference for the Ypresian-Lutetian transition in the southern North Sea Basin. Nevertheless, the identification of the base of the Lutetian in the Belgian Basin remains a moot point.
The base of the Lutetian has in the past been defined at the 'first occurrence datum' of the planktonic foraminiferal taxon Hantkenina at c. 48.6 Ma (e.g. Luterbacher et al. 2004), corresponding approximately to the base of calcareous nannofossil zone NP14 (Fig. 3[point 1]), but recent studies have shown this Hantkenina event to be significantly diachronous. In April 2011 the International Union of Geological Sciences ratified the proposal that the base of the Lutetian is defined by the Gorrondatxe GSSP (NW Spain) (Molina et al. 2011). This boundary is placed at the 167.85 m level, corresponding to the lowest occurrence (LO) of the nannofossil taxon Blackites inflatus (Fig. 3[point 3]). The latter is well known from many low and middle latitude outcrop and borehole sections worldwide (Southern USA: Siesser 1983; Russia: Shcherbinina 2000; off NW Australia: Shamrock 2010a,b; etc.), but seems to be missing at higher mid latitudes, such as the North Sea Basin (not mentioned by Perch-Nielsen 1971;Bigg 1982;Depret & Willems 1983;Steurbaut 1988;Verbeek et al. 1988;Varol 1998 andKöthe 2012; partially misinterpreted by Aubry 1983: figured specimens belong in fact to B. praeinflatus n. sp.). In situ specimens of B. inflatus have not been recorded at the Mont-des-Récollets during the present investigation. However, new intensive research on about 100 samples of the Brussel Sand Formation (Figs 12, 13; detailed by Steurbaut in a forthcoming study) has currently led to the identification of 13 specimens of Blackites inflatus on a total of about 100 000 specimens screened. Two specimens come from the Vlakte van de Raan borehole (one at 78.5 m and one at 74.5 m depth) in the extreme NW of Belgium and four from the Mol borehole in the northeast of Belgium; three were recorded in one out of twelf studied samples from the Isnes quarry (Fig. 20H, J), 10 km northwest of Namur, and the remaining four in one out of nine studied samples of the Gobertange quarry, about 30 km north of Isnes (Fig. 20I, K). At Isnes B. inflatus seem to co-occur with abundant Nummulites laevigatus in situ, whereas the latter was not reported from the Mol BH and only sporadically from Gobertange (Rutot 1893(Rutot , 1910, although not encountered by us. The extremely rare records of B. inflatus in Belgian sections seem to represent its most northern occurrence in NW Europe, which has up to now been thougth to correspond to the 'Glauconie Grossière s.s.' in the Paris Basin (Aubry 1983). Calcareous nannofossil investigation of the base of the historical Lutetian stratotype (e.g. at Margival and Gisors, Steurbaut 1988;Prémontré, Steurbaut et al. 2016b; Chaumont-en Vexin, in the present paper; see also Appendix 1) has revealed that B. inflatus is absent in the 'Chaumont-en-Vexin sands', representing the oldest Lutetian deposit of the Paris Basin (Ott d'Estevou et al. 2014; probably corresponding to level ('niveau') 1 of the Lower Lutetian of Merle 2008), and in he lowermost 2 to 3 meters of the overlying 'Glauconie Grossière s.s.' (as defined by Blondeau 1980). This interval seems to be devoid of Nummulites laevigatus (Fig. 17). The similarity in nannofossil succession between the Belgian Basin and the Paris Basin indicates that the base of the 'Chaumonten-Vexin sands' is coeval with the base of Unit A4 in the upper part of the Aalter Sand Formation. This is essentially based on the identification of the major nannofossil turnover in both areas. It also shows that the base of the 'Glauconie Grossière s.s.' is coeval with the base of the Brussel Sand Formation, as evidenced by the LO of Luminocanthus plenilutetiensis n. gen., n. sp. Furthermore, the nannofossil succession indicates that the 'Glauconie Grossière s.s.' is a highly condensed deposit, which was synchronously developing with the Brussel Formation, although with quite distinct sedimentation rates (2.5 m at Margival equates 5 m at Mont-des-Récollets and c. 40 m of infill in the Brussels megachannel). This is also true for the 'Chaumont-en-Vexin sands' with regard to Unit A4 (0.9 m at Chaumont-en-Vexin equates c. 3.15 m at Montdes-Récollets and 8.6 m in the Oedelem BH).
The stratigraphic significance of B. inflatus can be questioned, as, although it is quite common, this species is also irregularly distributed at low latitudes, especially in the lower part of its range. Its LO is located in the middle of chron C21r at the GSSP in the Gorrondatxe section (Payros et al. 2007), dated at 47.84 Ma (Payros et al. 2015, and at Agost (Larrasoaña et al. 2008: fig. 11 (Fig. 17).
If the internationally accepted Lutetian boundary criterium (LO of B. inflatus) should be applied to Belgium, it would mean that almost the entire Brussel Sand Formation is Ypresian in age, a conclusion that cannot be upheld neither historically nor paleontologically. Indeed, the Brussel Sand Formation (equating the "Bruxellian" of Dumont 1839, Rutot 1882; Gosselet 1883: 318; and many others), has always been considered to correspond to the lower part of the Lutetian (see Leriche 1939 for an overview). Moreover, the application of the boundary criterium would imply that the 'Chaumont-en-Vexin sands' and the lower part of the 'Glauconie Grossière s.s.' (as defined by Blondeau 1980), both constituting the base of the historical Lutetian stratotype, should also be of Ypresian age, and that is completely absurd.
Taking into account these historical considerations and the fact that the detection of the LO of B. inflatus is very problematic, even at low latitudes, it may appear that the LO of Discoaster sublodoensis would be a much more real-istic and reliable alternative criterion for defining the base of the Lutetian. This event has been identified worldwide (offshore Australia, Shamrock 2010b; Italy, Tori & Monechi 2013, Agnini et al. 2014Franceschi et al. 2015;Spain, e.g. Payros et al. 2015). Most importantly, it is recorded at the base of the historical Lutetian stratotype and is always associated, at least in NW Europe, with a major turnover in calcareous nannofossils. The latter is marked by many lowest occurrences (c. 10 taxa), which in turn coincide with the highest occurrence of Toweius occultatus, a widespread long ranging Ypresian taxon.
In conclusion, in order to meet the historical concept of the Lutetian Stage and to allow consistent identification throughout lower and middle latitudes it is proposed to amend the original definition of its lower boundary criterion (LO of B. inflatus) and to replace it by the LO of Discoaster sublodoensis. This nannofossil event is located in the middle of Chron C22n (Agnini et al. 2006(Agnini et al. , 2014Franceschi et al. 2015: supplementary material, (Fig. 17).

CONCLUSIONS
The key conclusions that emerge from the integrated lithostratigraphic and calcareous nannofossil investigation of the Mont-des-Récollets section, along with data from other sections in the southern North Sea Basin, are: 1) The entire suite of upper Ypresian and Lutetian formations and members, from the top of the Hyon Sand Formation upper NP12) to the base of the Maldegem Formation (Wemmel Sand Member, mid-NP15), is represented at the Mont-des-Récollets by lithofacies quite similar to that of their respective type areas in central Belgium. The upper Ypresian units, except for the Aalter Sand Formation, had never before been recorded in France, extending their spatial distribution considerably southwestward.
2) The lithofacies and thickness of the stratigraphic units in the Mont-des-Récollets outcrop indicate that deposition took place in a tectonically relatively quiescent area during late Ypresian and most of Lutetian times, in contrast to western Belgium, which underwent distinct uplift pulses of the Brabant Massif. However, this relative tectonic quiescence was temporarily disrupted at the Mont-des-Récollets, as the deposition of the interval with in situ Nummulites laevigatus (Unit B4) was preceded and followed by a substantial erosion phase. The lowermost erosion event, dated as mid-Biochron  NP14 (c. 48.00 Ma sensu Payros et al. 2015), destroyed most of the underlying Unit B3. This event seems to be local and linked to uplift of the Artois Anticline. The uppermost, dated as latest-Biochron NP14 (c. 46.50 Ma sensu Payros et al. 2015), almost totally dismantled the N. laevigatus bearing beds in the entire Belgian Basin, or only left some remnants, such as at the Mont-des-Récollets. These remnants represent the lower part of the N. laevigatus-bearing interval. The massively reworked large-sized N. laevigatus and N. laevigatus-bearing sandstone blocks at the base of the Lede Formation are unique relics of the upper part of this interval.
3) The subdivision of calcareous nannofossil Zone NP13 into three subzones, the major calcareous nannofossil turnover at the base of Zone NP14 (tentatively named BALCATevent) and the 9 bio-events within NP14 have been recorded throughout the Belgian Basin, including the Mont-des-Récollets, despite its impoverished assemblages. Several of these zonal boundaries and events have been identified in the Paris Basin, the Hampshire Basin (Steurbaut, in King 2016 and unpublished)  4) This study provides the first evidence that the 'Chaumont-en Vexin sands', base of the Lutetian historical stratotype, and Unit A4 of the Aalter Sand Formation, as well as the overlying 'Glauconie Grossière s.s.' and the Brussel Sand Formation (Bruxellian" sensu Dumont 1839), are time-equivalent deposits, the bases of which are coeval or almost coeval.
5) The similarity in calcareous nannofossil assemblages, the in situ specimens of Nummulites laevigatus and Campanile giganteum at the Mont-des-Récollets suggest that, after c. 2.3 Myr of isolation (latest Biochron NP12 and Biochron NP13), the direct north-south connection between the Belgian Basin and the Paris Basin was temporarely re-established during Biochron NP14 and during early Biochron NP15. and 6) The internationally accepted base-Lutetian boundary criterion (LO of Blackites inflatus) is difficult to apply in the North Sea Basin because of the extreme rarity of the marker taxon (not recorded in situ at the Mont-des-Récollets). Worst of all, its application would imply that the major part of the Brussel Sand Formation and, more important, the 'Chaumont-en-Vexin sands' and lower part of the overlying 'Glauconie Grossière s.s.', base of the historical Lutetian stratotype, should be of Ypresian age, which is a contradictio in terminis. To resolve this contradiction it is suggested to replace the original criterium (LO of B. inflatus) by the LO of Discoaster sublodoensis, which is recorded in many sections worldwide, including the base of the 'Chaumonten-Vexin sands' and the base of the Unit A4 of the Aalter Sand Formation. This bio-event is located within the middle of Chron C22n, about 800 kyr earlier than the LO of Turborotalia frontosa. This would also mean that the base of the Lutetian should be lowered down by about 130 meters at the Gorrondatxe GSSP, and consequently should range in age between 49. distributioN. -This taxon is recorded in several middle Eocene outcrop and borehole sections in Belgium (Steurbaut in Herman et al. 2001 andin Smith et al. 2004;Steurbaut et al. 2015). In the Mont-des-Récollets section it consistently ranges from the uppermost part of the Brussel Sand Formation (base of Unit B4) to the base of the Wemmel Sand Member. In terms of calcareous nannoplankton zones, its occurrence encompasses the upper part of NP14 and the lower part of NP15, as it appears to do in the Paris Basin (Aubry 1983(Aubry , 1986) and the Aquitaine Basin (Lin et al. 2017). In the Paris Basin it seems to recur in the Bartonian (Chavençon Marls: NP17 or NP18, Aubry 1983: pl. 8, figs 10-11), although this has not been observed in the other basins.

discussioN
B. stylifera differs from all up to now known Braarudosphaera taxa by its very high slightly tapering outline, resembling a pentagonal pyramid with truncated top, the margin sides of which are often slightly bowed inward (Fig. 18B). Both pentagonal ends can clearly be observed viewed from above when over-focusing and under-focusing with the microscope.  lates with lower middle NP14. A single specimen is known from the lower part of Unit A4 (top of the Aalter Sand Formation) in the Vlakte van de Raan borehole, one from the turritellid level at Whitecliff Bay (at 225 m above the base of the London Clay, dixit D. Curry), and two from the upper part of the 'Chaumont-en-Vexin sands' in the Paris Basin, all attributable to the base of NP14. In west Kazakhstan (e.g. Aktulagay), L. minutus seems to appear still earlier (within the middle of NP13) (King et al. 2013).

discussioN
This small nannolith with a box-like shape has a very distinctive morphology and optical interference pattern, allowing unambiguous and easy recognition. diagNosis. -This new genus is erected to include circular coccoliths consisting of a non-perforated, practically closed, central disk and a much smaller outer rim with many slightly bent elements (n > 30). The outer rim is always highly birefringent in cross-polarized light, the central disk sometimes (e.g. in the type species).

discussioN
Four species are currently included in this genus (in stratigraphical order): L. eolutetiensis n. gen., n. sp. from the lowermost Lutetian of the Belgian Basin (see below), L. plenilutetiensis n. gen., n. sp. from the lower middle Lutetian of the Belgian Basin and the Paris Basin (see below), L. hirsitus (Müller, 1970) n. comb. from the lower middle Oligocene of Belgium and Germany (Müller 1970;Steurbaut 1986) and L. hoerstgensis (Müller, 1970) n. comb from the upper middle Oligocene of Northern Germany (Müller 1970). Both Oligocene species have been synonymized with the upper Eocene species Cribrocentrum reticulatum (Gartner & Smith, 1967) (cf. Nannotax 3, http://www.mikrotax.org/Nannotax3/index. html), although erroneously, as these taxa are fundamentally different (e.g. in construction of distal shield and number of elements in the shields).
diagNosis. -Circular coccoliths consisting of two appressed shields and including a central disk covered by a series of radially oriented calcite laths and a smaller outer rim with numerous almost indistinct slightly curved elements. Only the outer rim is highly birefringent in cross-polarized light.
derivatio NoMiNis. -Refers to the total range of this species, which seems to be restricted to the earliest part of the Lutetian. diMeNsioNs. -Diameter = all around 9 µm (holotype: d = 9 µm).
distributioN. -Luminocanthus eolutetiensis n. gen., n. sp. is known from boreholes in the Belgian sector of the North Sea (SEWB and Vlakte van de Raan, see Appendix 1), where it seems to be retricted to the top of the Aalter Sand Formation (base of NP 14). Up to now it has not been recorded in coeval deposits from onshore Belgium.
descriptioN These medium-sized (d = c. 9 µm) rather thick subcircular forms consist of a prominent outer rim including numerous (probably c. 50) vaguely visible elements and an inner area with radially oriented elements. Both zones are sharply delimited in distal view. The outer rim is rather thick and exhibits high birefringence under crossed nicols, especially the outer rim of the proximal shield (Fig. 19Ba). The inner area is only faintly birefringent, although shows a clear extinction cross with laevogyre curvation in distal view. It widens and is somewhat blurry in the outer rim.

discussioN
This taxon differs from L. plenilutetiensis n. gen., n. sp. (Fig. 19G-J) by the configuration of the outer rim and the central area. The outer rim of L. eolutetiensis n. gen., n. sp. is much more elevated, smaller and sharply separated from the central area. This rim is also highly birefringent as in L. plenilutetiensis n. gen., n. sp., but contrarely to the latter, the central area remains only faintly illuminated in cross-polarized light. The extinction lines are rather sharp in the central area and become wider and vague in the outer rim.
diagNosis. -Circular coccoliths consisting of two closely appressed shields, including a central disk covered by a series of irregularly oriented calcite laths and a smaller outer rim with numerous conspicuous slightly curved elements. Both areas are highly birefringent in cross-polarized light.
derivatio NoMiNis. -Refers to its position within the Lutetian, as its lowest occurrence is at the base of the coarse-grained sand facies of the Lutetian (base of Brussel Sand Formation in Belgium, base of 'Glauconie grossière s.s.' in the Paris Basin), which overlie the basal Lutetian fine-grained glauconitic sand facies, as identified in many outcrop and borehole sections in Belgium (Unit A4) and at Chaumont-en-Vexin ('Chaumont-en-Vexin sands') in the Paris Basin. diMeNsioNs. -Diameter = 5.0 to 9.6 µm (holotype: d = 8.4 µm).
distributioN. -Luminocanthus plenilutetiensis n. gen., n. sp. is known from many Lutetian outcrop and borehole sections in the southern North Sea Basin, including the historical stratotype area (e.g. from the base and the upper part of the 'Glauconie Grossière s.s.' at Prémontré (Fig. 19I) and Margival respectively). In Belgium it is consistently present from the base (Unit B1) up to the top of the Brussel Sand Formation (e.g. Isnes quarry) and seems to be restricted to it. At the Mont-des-Récollets there is a major influx (also the first occurring specimens) at the base of Unit B2b. It has been recorded in Unit C2 (sample 50.8 m) of the Aktulagay section, attributed to NP14, although in very low numbers (Steurbaut in King et al. 2013).
descriptioN These rather small (generally around 7 µm in diameter) and thick circular coccoliths consist of two closely appressed shields.
The distal shield has a prominent outer rim including 32 to 38 elements, displaying laevogyre element curvature in distal view (Fig. 19G), and an inner concave area with less-well visible irregularly oriented elements. Both zones are not sharply delimited in distal view and show high birefringence under crossed nicols. The conspicuous swastika-like extinction cross has a laevogyre outline in the central area in distal view. It widens and is somewhat blurry in the outer rim.

discussioN
The extinction figure of the type species is quite similar to that of L. hirsitus (Müller, 1970) n. comb. from the lower Middle Oligocene of Belgium, suggesting a close relationship. However, coccoliths of the latter seem to have a much broader outer rim. The differences with L. eolutetiensis n. gen., n. sp. are discussed above.
Family discoasteraceae Tan

discussioN
In their investigation of middle Eocene calcareous nannofossil associations of two DSDP holes, Wei & Wise (1989b) separated the discoasters with high numbers of rays (16-27) and one prominent central stem from similar forms with fewer rays (10-15), known as Discoaster bifax Bukry, 1971, to which they previously have been attributed (e.g. in Perch-Nielsen 1985;Aubry 1986). The populations with high ray numbers were grouped into D. praebifax, which, in view of its morphological similarities and stratigraphic relationship, was suggested to be the ancestor of the latter. The number of rays in the Mont-des-Récollets specimens ranges between 20 and 22, evidencing the presence of D. praebifax only. distributioN. -Typical 5-rayed specimens.of Discoaster sublodoensis are recorded at the base of Unit A4 in the well-preserved assemblages of the Vlakte van de Raan borehole and at the base of the 'Chaumont-en-Vexin sands' in the Paris Basin, although always in very low numbers. They co-occur with numerous 6-rayed specimens with straight rays (considered herein as 'late form' of D. lodoensis) and other, although more rarely, 6-rayed forms with straight rays without ridges (D. strictus and alies). In the less well preserved assemblages of the Mont-des-Récollets the typical 5-rayed specimens seem to pop up only from the middle of Unit A4. D. sublodoensis remains quite rare within the totality of its range at the Mont-des-Récollets, although it is consistently found up to the top of Unit B3. It is extremely rare from the base of B4 onward and has not been observed in the base of the overlying Lede Sand Formation.  fig. 2). The latter are marked by a rather wide central area rapidly narrowing and extending into 5 or 6 (exceptionally 7) slender rays and by ray margins, which are slightly bowed inwards near the base of each ray. The area where the rays meet is not sharp and angular, but, on the contrary, distinctly rounded (see Fig. 18M). The rays of the North Sea Basin specimens have triangular free ends, with straight margins, which meet at clear angles (see Fig. 18I-K). The central disk, which is marked by a small low stem and low crests along the margin of the rays, ending at the interstices between the rays, is similar within the total population range of the species. Achuthan & Stradner, 1969 (Fig. 18O-R) Discoaster wemmelensis Achuthan & Stradner, 1969: 5, 6, text-fig. 2;pl. 4, figs 3-4. distributioN.

Discoaster wemmelensis
-D. wemmelensis is consistently recorded in Lutetian nannofossil assemblages worldwide, although generally in low concentrations (Perch-Nielsen 1985;Aubry 1986;Varol 1998;Tori & Monechi 2013;Franceschi et al. 2015). It occurs in very low numbers at the Mont-des-Récollets, from the base of the Brussel Sand Formation (its LO is at the base of Unit B1, lower NP14) up to the base of the Wemmel Sand Member (middle NP15). A few isolated specimens have been recorded slightly earlier at the top of the Aalter Sand Formation in boreholes north of the Mont-des-Récollets, but not at the Mont-des-Récollets itself (2 in the Vlakte van de Raan borehole and 1 specimen in the Oedelem borehole, a few meters above the base of NP14 as defined herein). This might not correspond to its total range, as in Belgium D. wemmelensis is known to occur throughout the Wemmel Sand Member (middle NP15) up to the Ursel Clay Member (base of NP16) (Steurbaut 1986).
paratypes. -1 from the "Chaumont-en-Vexin sands" at Chaumonten-Vexin (sample CHA11, 0.55 m above the base of the Lutetian (Fig. 18F descriptioN This circular slightly inflated body consists of a number of (generally three) peripheral concentric rings surrounding a central network, which are supported by 17 to 19 radially oriented pillars. This concentric structure, which resembles a spider's web, shows no birefringence under crossed nicols.

discussioN
This taxon has many similarities with Martiniaster fragilis, but instead of the 12 radially arranged peripheral ridges, it has a much larger number of radial pillars (generally between 17 and 19), which are less prominently developed. It appears to be also less globular.
Martiniaster fragilis (Martini, 1961) Loeblich & Tappan, 1963 ( Fig. 18N descriptioN This Trochoaster species is characterised by six symmetrically positioned protruding rays with strong median ridges along the rays. The ridges are connected with a central circular ring, which surrounds a triradiate central bridge. The entire body, also the areas between the centrally positioned ridges of the arms, is covered with numerous bumps. These flat interray areas show no furher ornamentation. Specimens of this taxon do not exhibit birefringence under crossed nicols.

discussioN
This taxon differs from all known Trochoaster taxa by its rugged and knobby outline and the configuration of the central ring, where the median ridges along the rays meet. It presents a vague resemblance with T. martinii from the Early Miocene of Germany because of the presence of six protruding rays. The latter is less knobby and presents some fundamental differences in configuration of the central structure and the interray areas. Genus

discussioN
Three rather long thin arms with bifurcating endings, which are lined with a narrow weblike fringe. diagNosis. -Three irregularly bifurcated arms interconnected with a fourth arm to form a pyramidal construction surrounded by a weblike structure.

Trochastrites pyramidalis
derivatio NoMiNis. -Refers to the outline of the solid inner structure of these globular bodies, representing a triangular pyramid. descriptioN These nannoliths present a pyramidal structure consisting of three irregularly bifurcated arms lying in a plane and interconnected with a fourth arm almost perpendicular to that plane (Fig. 20U). This structure is surrounded by a weblike network. The bifurcated part of the arms is quite large compared to the total length of the arms.

Toweius brusselensis
distributioN. -At the Mont-des-Récollets B. minusculus n. sp. ranges from the top of the Aalter Formation (upper part of Unit A4) up to the upper part of the Brussel Sand Formation (top Unit B3), although inconsistently and in low numbers. It has not been recorded in Unit B4, and rarely in the overlying Lede Sand Formation at this locality. It occurs in many outcrops of the Brussel Sand Formation in its type-area (Vossem, Neerijse, St-Lambrechts-Woluwe, Nederokkerzeel, etc.), essentially in Unit B3 and is known from Unit B4 in the Vlakte van de Raan BH (Steurbaut unpublished). This species is consistenly present in the 'Chaumont-en-Vexin sands' of the Paris Basin, attributable to the base of NP14. Hence, its range in the North Sea Basin seems to be restricted to NP14 and the lower part of NP15, with a major influx in the lower middle part of NP14 (Unit B3). B. minusculus n. sp. has been encountered in west Kazakhstan, also in low numbers, but appearing at a slightly lower position (middle of NP13) (King et al. 2013).

descriptioN
The rhabdoliths of this new taxon are tiny, ranging from 7.5 to 10.4 µm in length. They consist of a slender tapering stem, without a collar, surmounting a thin basal plate. The stem is slightly inflated in its upper part, reaching its maximum thickness at about 1/5 of its length, measured from the rounded distal tip. Its surface is rugose. The basal plate consists of 1 layer of radially oriented elements.

discussioN
The presence of an elongated stem, rising vertically from a rather small and thin basal plate, and its optical characteristics (typical vertical extinction line and interference colours of the stem) allow this taxon to be attributed to the genus Blackites. It is distinct from all other described Blackites taxa by its tiny rhabdoliths (generally around 8.5 µm in length, maximum width c. 1 µm). Its silhouette resembles that of B. perlongus, although the latter is twice as long (17- diagNosis. -Slightly and somewhat asymmetrically inflated robust rhabdoliths, with faintly rugose surface but without a basal enlargement or collar, of which the maximum width of the stem is much less than the width of the more or less complexly structured basal plate. derivatio NoMiNis. -Refers to the outline of these rhabdoliths, which is inflated, although only slightly and asymmetrically compared to that of B. inflatus, to which it bears a certain resemblance. descriptioN This species is marked by weakly inflated half-long rhabdoliths with slightly rugose surface and which distally gradually taper to a sharp tip (length = 22 to 29 µm, about 6 times greater than the diameter of the base, which is about 3.3 to 5 µm). The stem starts thickening distally, although only slightly and somewhat asymmetrically, at about 1/6 of its length (not at the base, where the width is only 80% of the maximum width), to reach its greatest diameter at about 1/3 of its length. The maximum width of the stem (c. 2.8 µm) measures about 14% of the total length of the rhabdolith, and is always substantially smaller than the diameter of the basal plate. The latter is complexly structured and seems to consist of two cycles of elements (Fig. 20G)

discussioN
This medium-sized rhabdolith (c. 11 µm length) is marked by a very wide basal plate (c. 4.8 µm) and a lensiform stem, which maximum width is located in the upper half of the stem. The latter, which presents a basal collar, seems to be thin-walled (Fig. 20L) and rugose towards the top. These rhabdoliths bear great resemblance with those of B. gladius (Varol 1989: pl. 4, fig. 5;Bown 2005: pl. 21, figs 10-14). However, they differ by certain features of the stem, which is relatively longer and more slender in B. aff. gladius, especially in the top part, as well as by the position of its maximum width. The latter is less distal compared to its position in B. gladius. hoLotype. -Fig. 20N (IRSNB b7154) (negatives stored in the collections of the RBINS).
distributioN. -This species has a rather restricted range. It occurs in the top of the Brussel Sand Formation (Unit B4) and in the Lede Sand Formation at Mont-des-Récollets, and in several outcrops of the Lede Sand Formation (Balegem, Oosterzele) and the base of the overlying Wemmel Sand Member (Oedelem borehole). Its range seems thus to be restricted to the top of NP14 and lower NP15.
descriptioN These sphenoliths are very tiny, consisting of a conical proximal column and a series of blocky lateral elements, which seem to widen distally, producing the typical hourglass shape. The latter is also easily observed in cross polarized light (see discussion below). Both the proximal column and the area with the lateral elements are of almost equal height (each c. 2.5 µm). Half long diverging spines (maximum 1.5 times as long as the sphenolith's body) emerge from each of the four corners.
discussioN Sphenolithus quadricornutus n. sp. bears a striking resemblance to Sphenolithus quadrispinatus, described from the Late Miocene (Tortonian, NN10) of the southern Atlantic (Perch-Nielsen 1980), especially in its dimensions and the length and the position of the spines. They essentially differ in general outline, which is cylindrical in S. quadrispinatus, but hourglass-shaped in S. quadricornutus n. sp. This is also clearly observed in cross-polarized light when seen at 0°. In this position S. quadrispinatus is highly birefringent, consisting of four almost equal rectangular blocks (height is greater than width), separated by extinction lines with cross pattern.
GEODIVERSITAS • 2021 • 43 (11) S. quadricornutus n. sp. is also strongly birefringent in that position, but presents an x-shaped interference figure, of which the four bright parts are rather rhombic instead of rectangular and separated by extinction lines with cross pattern. Viewed at 45° this cross pattern turns into an x-pattern. diagNosis. -Rather large sphenoliths, consisting of an almost cubical lower portion, including the proximal shield and lateral elements, and a conspicuous apical spine with particular optical characteristics, made up of a solid centre and two diverging sword-like blades, which remained joined together over more than half of their length, before diverging distally and finally ending in a long distal threadlike structure. Both blades diverge at an angle of about 70°.
distributioN. -This species has a very restricted range, both in time and in space. Up to now it has only been recorded at Montdes-Récollets, and only in the top of the Aalter Sand Formation (upper part of Unit A4) and the base of the Brussel Sand Formation (Units B1 and B2) (lower middle of NP14).
descriptioN These sphenoliths are robust and rather large, ranging from 11.2 to 14.0 µm in height. Their proximal portion consists of an irregular lattice, almost quadrangular in lateral view (width slightly larger than height), including a basal layer of radial elements (proximal cycle) surmounted by layers of lateral elements (lateral cycle). The apical spine consists of a solid centre and two diverging sword-like blades. These blades remain joined together over more than half of their length, before diverging at an angle of c. 70°. The top of each blade is tapering and ends in a thin and moderately long distal threadlike structure. These structures, as well as the top of the blades, are fragile and are often abraded or broken off. The outline of these sphenoliths is unique, as well as their optical behaviour. Viewed between crossed nicols at 0° (in position parallel to the polarization directions) the entire sphenolith shows maximum birefringence, with a conspicuous extinction cross in the proximal portion and a vertical extinction line in the apical spine, separating the two halves. Maximum birefringence occurs every 90°. When viewed at 45° the extinction lines in the proximal portion have shifted from a cross-pattern into an x-pattern, whereas only the central part of the apical spine remains strongly birefringent and the blades only faintly. At 32° from maximum birefringence one of the blades becomes completely extinct (also most of its centre). The other blade fades out at c. 26° further clockwise (at c. 58° from the initial starting point).

discussioN
The outline, structure and optical behaviour of the proximal portion of this new Sphenolithus taxon is very similar to that of Sphenolithus radians (Fig. 19Q). Both differ by the outline of the apical spine, which consist of 2 diverging blades in S. recolletensis n. sp. The outline of the apical spine of the latter presents some similarities with these of S. furcatolithoides and S. perpendicularis (see Shamrock 2010a: pl. 1), but clearly differs from these by its massive strongly birefringent candle-shaped form, when viewed at 45°. The angle between the 2 diverging blades of the apical spine (c. 70°) is intermediate to that of S. furcatolithoides (c. 35°) (see Bown 2005: pl. 45) and that of S. perpendicularis (c. 96°) (Fig. 20R). The latter, which has been discovered in the middle Lutetian offshore W Australia (lower part of Zone NP15, restricted to Zone CP13a) (Shamrock 2010a, b), has not been identified at the Mont-des-Récollets. However, a single specimen has been encountered in the middle of the Lede Sand Formation at Balegem (Fig. 20R). This is the first record of S. perpendicularis in the Belgian Basin. Bukry, 1971 (Fig. 19P) Sphenolithus spiniger Bukry, 1971: 321, pl. 6, figs 10-12; pl. 7, figs 1-2.

Sphenolithus spiniger
distributioN. -The lowest occurrence of typical Sphenolithus spiniger (as defined below) is in the lower part of the Brussel Sand Formation (at Mont-des-Récollets in Unit B2). It is known from many outcrop and borehole sections of the Brussel Sand Formation and increases in numbers during the Sphenolithus bloom, as recorded in the Diegem and Zaventem outcrop sections (Steurbaut, detailed in forthcoming study). It occurs rarely in the Lede Sand Formation.

discussioN
This sphenolith differs from the other Sphenolithus taxa described herein by its small triangular outline and very distinctive extinction pattern under crossed nicols. Viewed parallel to the polarisation directions it consists of 4 lobes separated by an extinction cross. The 2 proximal lobes are large and oval; the upper 2, which correspond to the lateral elements and spine, are much smaller and triangular. The extinction cross is much broader at its proximal end, so that the separation between the two lower lobes increases in proximal direction. Forms with a similar outline, but with a clearly different extinction pattern, are attributed to Sphenolithus aff. spiniger, known from the upper part of the Aalter Sand Formation and the Brussel Sand Formation. In these foms the lower 2 lobes are also triangular and the extinction cross is not enlarged proximally. diagNosis. -Large, irregular network of calcite laths, joining at the centre in a central protruding x-shaped wedge.
derivatio NoMiNis. -The name is a combination of the Christian names Joachim and Celine, introduced to commemorate and celebrate the wedding day of these two lovely persons. descriptioN This taxon is composed of an irregular complex network of several long calcite laths (up to 7), meeting at different angles in the centre of a more or less cubic structure. The dimensions are highly variable, as the different laths are often broken off. The longest diagonal reaches up to 17.6 µm, the second is generally slightly shorter. The centre is marked by an x-shaped protruding wedge. All calcite laths are strongly birefringent in cross-polarized light.
discussioN Similar lathlike networks with central X-shaped to diamond-shaped wedges were described from upper Ypresian deposits of the southern North Sea Basin (Steurbaut 2011). Nannoturba joceliniae n. sp. differs from all previously described Nannoturba species (N. robusta, N. spinosa and N. jolotteana) by a reduced number (up to 7) of long, almost completely free, irregularly distributed laths, which only meet at the centre in an x-shaped to diamond-shaped wedge. In all others taxa the laths are joined over most of their length with different configurations, according to the species (N. robusta with quadrangular disposition and laths meeting at 90°; N. spinosa with triangular disposition with laths meeting at 120°; N. jolotteana (Fig. 19A) resembling a flattened octahedron, with rhombic cross-section). It differs from specimens described as Nannotetrina nitida by Aubry (1983: pl. 5, figs 13-14), as well as from the latter's holotype (originally described as Tetralithus nitidus by Martini 1961: pl. 1, fig. 5, pl. 4, fig. 41), by its large number of different sized laths. In fact, the specimens figured by Aubry also seem to belong to the genus Nannoturba, but probably represent another new species, different from Tetralithus nitidus Martini, 1961. GEODIVERSITAS • 2021 unpublished data and samples from Chaumont-en-Vexin and Fleury-la-Rivière, and to Dr J. Tyler (honorary deputy director of the Smithsonian Institute, Washington) for his careful review of the English language. Thomas Goovaerts (RBINS) and Stéphane Berton (RBINS) are responsible for cutting and polishing of the sandstone samples. Drs Florence Quesnel (BRGM, France) and Monechi Simonetta are thanked for their constructive comments and suggestions, improving the manuscript.