Micromorphology and Site Formation at Hohle Fels Cave , Swabian Jura , Germany

Hohle Fels Cave near Schelklingen in the Swabian Jura area of southwestern Germany, contains an occupational and sedimentological record that spans at least the last 36,000 yeats and includes Aurignacian, Gravettian, and Magdalenian prehistoric occupations. The sediments were subjecred to detailed micromorphological analysis coupled with microanalytical data using electron microprobe and FTIR techniques. The results show that much of the sediment has been de­ rived from the interior of the cave where the finer ma­ trix was partially phosphatized, likely derived from bear habitation. Moreover, the sediment has been subjected to cryoturbation and ice lensing under cold and damp conditions. These cold-related features become increas­ ingly well developed in the Gravettian and Magdale­ nian layers reflecting more marked cooling during these periods. There is no evidence for the occupation of the cave by humans or cave bears during the Last Glacial Maximum. These geoarchaeological observations can be used test hypotheses about paleoclimate and human behavior developed using independent lines of evidence provided from botanical, faunal and archaeological ma­ terials. Some of the methods employed in this research have not previously been used to study the caves of the Swabian Jura and provide key new insights into the ar­ chaeological and natural history of the region. 'Anschrift der Verfasser: P. GOLDBERG, K. M E ­ LIGNE, C. DAYTON: Department of Archaeology, Boston University, 675 Commonwealth Avenue, Boston Massachusetts 02215, U.S.A. E-mail: paulberg@bu.edu; N . CONARD, S. SCHIEGL: Institut für Urund Frühgeschichte und Archäologie des Mittelalters der Universität Tübingen, Abteilung für Ältere Urgeschichte und Quartärökologie, Schloss Hohentübingen, D-72070 Tübingen Zusammenfassung: Die Höhle Hohle Fels liegt auf der Schwäbischen Alb bei Schelklingen und beinhaltet eine stratigraphische Folge, die mindestens 36.000 Jahre zurück geht und Aurignacien-, Gravettienund Magdalenien-Fundschichten beinhaltet. Die Sedimen­ te vom Hohle Fels wurden mit mikromorphologischen Analysen in Kombination mit Elektronmikroprobe und FTIR-Analysen untersucht. Die Ergebnisse zeigen, dass die Sedimente aus dem inneren Bereich der Höhle stammen und das die feine Matrix in Zusammenhang mit der Nutzung der Höhle durch Bären eine starke Phosphatenanreicherung erlebt hat. Mikrostruktu­ ren belegen kalte und feuchte klimatische Phasen, chararkterisiert durch Kryoturbation und Eislinsen. Diese Merkmale sind in den Gravettienund Magdalenien-Schichten stärker entwickelt und sprechen für kühle Bedingungen während dieser Perioden. Während des letzten Kältemaximums fehlen Hinweise für die Nutzung der Höhle durch Menschen und Höhlen­ bären. Diese Beobachtungen können als Grundlage dienen, um Hypothesen über das Paläoklima und über menschliches Verhalten im Paläolithikum, die anhand botanischer, faunistischer und archäologischer Daten entwickelt wurden, zu prüfen. Einige der Methoden dieser Untersuchungen wurden zum ersten Mal in den Höhlen der Schwäbischen Alb eingesetzt und lieferten viel versprechende Einblicke in die Archäologie und die naturhistorische Enrwicklung der Region.


Site Setting and Stratigraphy
Hohle Fels Cave is situated on the eastern ex tension of the Swabian Jura about 20 km west of Ulm (Figures 1 and 2).The cave is formed within an Upper Jurassic (Malm) limestone reef and occurs at about 534 m above sea level (asi) in the valley of the Ach River, a tributary of the Danube.The Ach Valley was cut by the Danube before it left its bed shortly during the Riss gla cial maximum (VIELINGER 1986).Today the Ach Rivet is about 3 m wide and flows to the east into the blau river.Prior to the Riss Glacial Peri od the valley bottom was up to 40 m lower than today and much narrower; it has subsequently been filled with gravel and sand.The cave is located on the southeast side of the valley, about 7 m above the current valley bottom which here is about 350 m wide.During the terminal Pleis tocene the valley bottom was ca.5-10 m lower than today's position (WAGNER 1979).
Hohle Fels is one of the largest caves in the re-Fig.1: Location Map of Hohle Fels Cave in the Swabian Jura.
gion with a large cave hall that has a maximum height of 12 m and covers an area of 500 m 2 .
This latest work has revealed a sequence of ebou-/w-rich silty and clayey deposits, some rich in charcoal and burnt bone.In addition, a number of archaeological horizons were recognized that range from Aurignacian strata at the base through Gravettian and Magdalenian layers at the top (Figures 3, 4; Table 1).
The stratigraphic sequence at Hohle Fels can be considered from the standpoints of both litho- on parameters such as color, texture, and internal organization.These are described in Table 1.
The archaeological horizons (AH) or Units, on the other hand, are determined on the basis of the assemblages of artifactual material.In the case of Hohle Fels, lithological and archaeological boundaries are strongly linked.At a more general level of analysis four main stratigraphic units (A-D) have been defined for operational reasons (CONARD & UERPMANN 1999; Table 1).
Each of these units is discussed from bottom to top as follows: Unit  Notes on descriptions: 1} These are provisional descriptions based on unpublished work by J. Hahn (12.09.1994);Waiblinger (1997); N.J.Conard and T. Prindiville (02.09.1997); and P. Russell and N.J.Conard (09.98).The geological and archaeological horizons (GH and AH) serve as preliminary field descriptions and are subject to modification during future excavations.
2) Colors were determined using the Munsell Soil Color Chart on moist samples, both in the cave under artificial light conditions and outside the cave in natural light.
3) Differentiation of individual sediment units on the basis of their calcium carbonate contents was not possible due to the large number of limestone clasts.
of archaeofauna has been systematically burnt for fuel.Fragments of bone of large game, such as mammoth and rhinoceros are also present.
Large, medium and small game, such as fox Horse and reindeer were the most important game.In addition, the layers contain bones of many other mammals, among which fox and hare are abundant.
3 Laboratory Methods

Samples and sample preparation
Cave sediments are typically complex mixtures of cultural and geological materials that are commonly diagenetically altered (WEENER et al. 1995).We thus chose a micromorphological approach that includes the study of undisturbed samples by light microscopy coupled with SEM study and elemental analysis of the in tact sedi ments.
A set of thirteen large samples of undisturbed blocks of sediment was collected along the western profile of the excavated entrance section (Figures 3 and 4).An additional sample (no. 14) was taken in excavation square 77, about 1 m away from the profile.The cave sediments characteristically contain major amounts of limestone gravel -mm to several dm in size -derived from the roof and walls of the cave.
Sampling such stony deposits is difficult, and it was not possible to use Kubiena boxes.Instead, we jacketed large areas of sediment exposed in Profile 2 with burlap dipped in plaster of Paris.
After the exposed part of the plaster jackets hardened, the block was carefully removed from the profile and the rear portion was similarly covered with plaster-impregnated burlap.
In the laboratory several 1-2 cm 2 large windows were cut through the plaster prior to 6 weeks Two series of thin section were produced, large sections of sizes of 7 cm x 10 cm for transmitted light microscopy and smaller ones (3.5 cm x 2.5 cm) with highly polished surfaces for scanning electron microscopy.
In addition to the sample blocks, a second se ries of loose sediment were collected from cor responding localities for infrared spectroscopy.
The latter were dried in a heating oven for sev eral days at 60° C.  1 and 2; Figure 4).Unit B is culturally sterile.The subdivisions within Unit A are Lay ers lk and 0c and date to the Magdalenian and Holocene, respectively (Table 1; Figure 3).

Coarse Fraction
The coarse elements, defined as grains of sandsize and larger (60 pm), consist of anthropo genic, geogenic, and biogenic components.The anthropogenic elements are burnt bone, some of the unburnt bone (if modified by humans), pos sibly coprolites, and charcoal.Geogenic com ponents include limestone clasts and calcareous "Ornaments" which crystallized on the roof and walls and were then detached.Two components of the coarse fraction could be attributed to both human and other biological agents: coprolites(/ phosphatic grains?) and unburnt bone.A few other minor constituents were observed in some of the thin sections: quartz grains, iron particles, and the coprolites mentioned above.
Bone, either burnt or unburnt, is nearly ubiq uitous throughout the sediments, except in the culturally sterile Unit B (Table 2).While we found unburnt bone in all the Gravettian and Magdalenian layers, burnt bone was present in significant quantities only in layers 3b, 3bt, and 3cf of the Gravettian period.The size of the bone fragments ranges from unidentifiable splinters (as small as 20 urn) to cm-long pieces.
Several tooth fragments are also present and occur in the matrix or are incorporated within aggregates (Figure 5).7b).Phos phatization of calcareous grains does not appear to follow a specific stratigraphic pattern, and it may be that phosphatization occurs in localized areas throughout the entire sequence.Ambercolored phosphatic grains of indeterminate, but likely biogenic, origin occur in all samples (Table 2; Figures 8a, 8b).

Fine Fraction
The fine fraction consists of a yellow-brown to brown calcareous and locally phosphatic clay.
Rare silt size inclusions of quartz grains, ambercolored phosphatic grains, and organic material -as well as silt-size fragments of the bone and charcoal described above -are embedded within the clay.
Although this calcareous clay is present throughout the sequence, the matrix in Layer 3c clearly differs from that of any other layers Etching of the calcite and replacement by phosphate creates an tooth-like pattern on the edge of the grain.PPL; width of view, ca.650 pm.
Fig. 8a: Microphotograph of slightly darker, amber-colored phosphatic grains in sample HF-8 (Layer 3b).Note their isotropic nature and the siltsize inclusions of quartz in XPL.They appear to represent phosphatic clays that have been reworked by cryoturbation or movement of material from within the main hall of the cave.PPL; width of view, ca.650 pm.As discussed below, the dip of the deposits from the interior of the cave to the exterior clearly in dicates that the clay is derived from the interior of the cave.Moreover, the clay appears to be ultimately derived from soil materials that were Fig. 9a: Sample HF-3C (Layer 3c).Microphoto graph of a domain of decalcified clay within the generally calcareous matrix.While the matrix appears undifferentiated in plane-polarized light, crosspolarized examination reveals distinct domains of isotropic, decalcified clay.This decalcification of the matrix occurs only within Layer 3c of the Gravettian complex.PPL; width of view, ca.3.2 mm.The individual rounded aggregares are clearly separated and non compacted.PPL; width of view, ca.3.2 mm.

Fabric
As discussed above, the fine fraction is generally heterogeneous calcareous clay whose fabrics can be described as combinations of a small number of solid patterning (matrix) and non-solid pat terning (voids).We divide the solid patterning into four modes of organization and the nonsolid patterning into three modes.

Solid patterning:
1) This is simply a loose non-ag gregated mix of clay and inclusions (Figure 10).
2) The second type is characterized by clearly defined individual rounded aggregates or ,loose rounded aggregates' (Figure 11).These indi vidual aggregates in some cases have a core com posed of a coarse component, such as a fragment of bone, charcoal, or lithic material (see Figure 5).3) In this case, the aggregates are compressed into more massive compound subangular ,macroaggregates' (Figure 12).In such instances, the overall microstructure of the matrix is fissured or subangular blocky, and pedality of the material is evident.4) Finally, the matrix also occurs as a massive, non-aggregated form, where the clay is relatively homogeneous compared to the type 3, above (Figure 13).unpatterned, displaying irregular interstitial (packing) voids between singular and compound aggregates (Figure 11).3) This last form is char acterized by fissures and cracks (Figure 12).
In all thin sections examined, platy voids are generally encountered either with the loose nonaggregated matrix (Figure 10) or with the loose rounded aggregates (Figure 14; Table 2).Van Vliet-Lanoe ( VAN VLIET-LANOE 1985) reports that platy structure due to ice lensing is best ex pressed in homogenous sandy substrates.In the case of Hohle Fels the heterogeneity of the sedi ments and the presence of large coarse elements -primarily and ultimately products of roof fall -makes the ice lensing structure less evident.
However, the platy organization is still visible in localized domains.
The interstitial void structure is mostly an at tribute of the loose rounded aggregates (Figure 11), and both patterns are characteristic of the effect of cryoturbation ( VAN VLIET-LANOE 1982, 1987, 1991, 1998).Fissures and cracks, on the other hand, are associated with both the massive subangular macroaggregates (Figure 12) and with the massive non-aggregated matrix (Figure 13).
In this context of extreme heterogeneity, the dif ferent types of fabric tend to coexist within the  same thin section commonly being found ad jacent to one another (Figure 15).Within each sample, however, it is possible to identify general fabric trends, as discussed below.

Textural features
Three varieties of textural or post-depositional features can be observed in thin section.Each type entails finely bedded matrix material -clay, organic matter, and fine lithic, bone and char coal fragments -redistributed around coarse elements, such as aggregates or limestone clasts.
The principal types of textural features include 1) coatings around the entire margins of coarse grains, 2) cappings on upper surfaces (Figure 16) and pendents on lower surfaces.Most coatings and cappings are poorly sorted, although they are finely bedded, except for overall coarser cappings in sample HF-5, which are constituted of sorted, inversely graded matrix material (Figure 17).
Thick cappings are characteristic throughout the Gravettian 3b unit (Table 2).The pendents are rare, as they were observed only in samples HF-12 and HF-13, both located on the tight side of the profile, away from the central axis where most of the samples were taken.The second crystalline feature appears creamy white in PPL, and white to gray under XPL and is a calcite precipitate (identified by FTIR and SEM) (Figures 19a,19b,19c).Although Fig. 18: Microphotograph of a microlayered calca reous ,ornament'.Note the finely bedded precipitate with irregular form .The precipitate was originally arranged around a limestone clast (see remains fragment of oolitic limestone, bottom left).It either represents a broken pendent formed on the underside of rock resting on the floor of the cave, or more likely was probably detached from the walls or ceiling of the cave.Sample HF-7B.PPL; width of view, ca.3.2 mm.most examples of this crystalline feature have irregular shapes, a few are well rounded.The mineralogy is consistent with their moist, pasty feel in the field, similar to that of moonmilk (GILLIESON 1997: 117, 122,126).These crys talline features occur in the upper part of the stratigraphic section, mostly in Layers Is, 3as and lk (Figure 4; Tables 1 and 2).

Discussion
In the field, minor lithostratigraphic differences are evident in the exposed profile, mostly distin guishable by differing amounts of roof fall and slight color changes in the matrix.In general, the sediments appear to be similar (see Table 1 and Figure 4).However, examination of the thin sections provides a much greater level of resolution, revealing nuanced differences that provide significant insight into the geological history of the cave deposits.The overall pattern Fig. 19 a+b: Sample HF-7B (Layer lk/Oc) is from the upper part of the Hohle Fels profile, which is typically lighter in color.The microphotograph illustrates irregularly shaped, sub founded precipitates of calcite, which are creamy white in PPL and white and gray under XPL.The surrounding fabric includes subangular macroaggregates associated with massive structure, fissures, and cracks, a) PPL, b) XPL; width of view, ca.3.2 mm. of the results is presented in Table 2.These pat terns reflect changes in climatic conditions, an thropogenic activity, and other depositional and post-depositional processes discussed below.

Climatic conditions
The profile from which these samples were col lected lies within the entrance passageway to the main chamber of the cave, and is approximately 20 m from the actual cave entrance (Figures 3   and 4).Nevertheless, the textural, composi tional and fabric data demonstrate that climatic conditions -and their fluctuations -that existed outside the cave can be inferred from trends in these sediments.The current floor of the cave is about 2 m higher than it would have been in Gravettian times due to subsequent deposi tion of geological and anthropogenic material.19a, b), but also includes grains of quartz (high Si peak) and bone (peak of P and Ca).Iron is likely tied to the clay matrix.ments were exposed to weathering under acidic conditions, presumably associated with the de cay of organic materials (e.g., vegetal material, guano).In contrast, the overlying sediments of Layer 3b, represented by samples HF-11, HF-5, HF-8 and HF-6, overall have a much "fresher" appearance microscopically.While these upper sediments also contain a significant phosphatic fine component, the phosphate is primarily in the form of sand-size amber grains that ate in corporated into fabric structures (Figure 12).PissARTetal. 1988aPissARTetal. ,. 1988b; VAN VLIET-LANOE 1991).In the overlying Layer 3b, however, the matrix is arranged in loose rounded aggregates clearly separated by interstitial/packing voids, that in certain cases is superimposed by a platy microstructure (Figures 11 and 14).Such rounded aggregates and the general absence of ice lensing features point to cryoturbation (VAN VLIET-LANOE 1982, 1987, 1991, 1998) 2).These two features are entirely absent from the older, Gravettian lay ers, and the question of their unique presence in the upper part of the sequence of the cave is intriguing and problematic.One interpretation is that they are related to a change of the overall climatic regime of the area.As such, dripping due to wetter and warmer (i.e., less freezing) climate could be responsible for the formation of both crystalline structures.The Magdalenian predates the warmer Bölling and Alleröd peri ods which are then followed by the Younger Dryas cold period.However, the paragenesis of both of these secondary features is still under study.In sum, it is necessary to note that there is a substantial temporal gap at the LGM and the early dates for the Gravettian, coupled with a relatively long period of non-human-occupa tion of the cave until ca.13ka b.p.

Anthropogenic activity
In the field, two main phases of Paleolithic oc cupation of the cave were identified between  The micromorphological investigation of Layer 3cf concluded that there was no evidence of in situ burning and that the layer of burnt bone had been most likely deposited by dumping.
Nevertheless, the extreme rarity of charcoal within the burnt bone Layer 3cf was confirmed.
Our micromorphological examination of the Gravettian complex as a whole seems to confirm

'
Anschrift der Verfasser: P. GOLDBERG, K. ME LIGNE, C. DAYTON: Department of Archaeology, Boston University, 675 Commonwealth Avenue, Boston Massachusetts 02215, U.S.A. E-mail: paulberg@bu.edu;N. CONARD, S. SCHIEGL: Institut für Ur-und Frühgeschichte und Archäologie des Mittelalters der Universität Tübingen, Abteilung für Ältere Urgeschichte und Quartärökologie, Schloss Hohentübingen, D-72070 Tübingen Zusammenfassung: Die Höhle Hohle Fels liegt auf der Schwäbischen Alb bei Schelklingen und beinhaltet eine stratigraphische Folge, die mindestens 36.000Jahre zurück geht und Aurignacien-, Gravettien-und Magdalenien-Fundschichten beinhaltet.Die Sedimen te vom Hohle Fels wurden mit mikromorphologischen Analysen in Kombination mit Elektronmikroprobe und FTIR-Analysen untersucht.Die Ergebnisse zeigen, dass die Sedimente aus dem inneren Bereich der Höhle stammen und das die feine Matrix in Zusammenhang mit der Nutzung der Höhle durch Bären eine starke Phosphatenanreicherung erlebt hat.Mikrostruktu ren belegen kalte und feuchte klimatische Phasen, chararkterisiert durch Kryoturbation und Eislinsen.Diese Merkmale sind in den Gravettien-und Magdalenien-Schichten stärker entwickelt und sprechen für kühle Bedingungen während dieser Perioden.Während des letzten Kältemaximums fehlen Hinweise für die Nutzung der Höhle durch Menschen und Höhlen bären.Diese Beobachtungen können als Grundlage dienen, um Hypothesen über das Paläoklima und über menschliches Verhalten im Paläolithikum, die anhand botanischer, faunistischer und archäologischer Daten entwickelt wurden, zu prüfen.Einige der Methoden dieser Untersuchungen wurden zum ersten Mal in den Höhlen der Schwäbischen Alb eingesetzt und lieferten viel versprechende Einblicke in die Archäologie und die naturhistorische Enrwicklung der Region. 1 Introduction Paleolithic excavations in the caves of the Swabian Jura have been conducted since the 1860s.This rich research tradition including the work of Oscar Fraas, R. R. Schmidt, Gus tav Riek and others has its intellectual and methodological roots in the geosciences, and since its origins in the early 20th Cen tury, the Department of Early Prehistory and Quaternary Ecology has had strong links to the geosciences of the University of Tübin gen.Despite the long and successful research tradition in geoarchaeology in Tübingen, which includes recent work by J. Hahn, (HAHN 1988), H. LAVII.EE (LAVII.EE & HAHN 1981) and I. CAMPEN (1987), it was not until the late 1990s that contemporary methods including micromorphological studies and microanalytical techniques, such as electron microprobe and FTIR have been applied to answer archaeological and paleoenvironmental questions related to the Paleolithic cave excavations in the Swabian Jura.With this paper we present a series of results using new methods to address research questions that range from site formation to environmental change, from the current excavations at Hoh le Fels near Schelklingen in the Ach Valley of southwestern Germany.
the Paleolithic from Hohle Fels in his classic monograph, "Die diluviale Vorzeit Deutschlands".From 1958-1960 G. Matschak and G. Riek conducted excavations in several parts of the cave, most notably in a large niche on the left side of the passage leading to the main hall of the cave.The excavators never published their results, and C. Saier (SAIER 1994) was the first to systematically study the finds from these excavations.Between 1977-79 and 1987-1996 Joachim Hahn renewed excavations in the niche in the hopes of gaining a stratigraphic section for comparison with the profile from nearby Geißenklösterle (HAHN 1988).Archaeological, palaeontological and geological research results from Hohle Fels have been published in numerous articles and theses (BI.UMENTRITT & HAHN
Fig. 3: Profile 2 showing stratigraphy with archaeological units and geological layers, as well as sampling locations.The inset map shows the deep interior chamber and the location of excavated area with exposures of sediments discussed in the text.The grid lines are 1 m apart.
and hare played a prominent part of the human economy -including use of hides, raw material and food.Unit B (archaeologically sterile) -Unit B con sists of two layers (Is and 3as), which together reach a maximum thickness around 70 cm next to the wall of the cave in the northern part of the excavation (Figure 4); these deposits thin out toward the interior of the niche.In con trast to the other strata, the horizons of unit B are light colored.This feature is due to high proportions of rubble-and sand-sized lime stone, and low amounts of clay, especially in the upper horizon.The B-complex is archaeologi cally nearly sterile and lacks datable material.The rare artifacts recovered from this complex are apparently not in situ and were likely in corporated into the sediment by cryoturbation.Consistent with its sterile nature, its sedimentological composition, and the radiocarbon dates from the over-and underlying deposits, the B-complex can be reasonably correlated with the Last Glacial Maximum.No clear evidence of occupation has been discovered so far in the Swabian Jura during this interval.Unit A (the Magdalenian) -The Magdalenian complex has been excavated over an area of 30 m 2 and has a thickness of 20-60 cm.Much of the sediments has been modified by cryotur bation inhibiting the establishment of a fine stratigraphy (CONARD & FLOSS 2001).The upper layers also contain diverse finds from the Holocene, such as pottery, demonstrating that considerable mixing has occurred.The Magdalenian complex dates to ca. 13,000 years BP.The lithic artifacts from the Magdalenian deposits include mostly blades and bladelets, although scrapers, burins, borer, backed knives and backed points have also been uncovered.They are produced mainly from local Jurassic chert and radiolarite, although non-local raw materials are also present (BURKERT & FLOSS in press).Numerous bone and antler tools, such as pro jectile points, one double-rowed harpoon and several needles also occur.The Magdalenian assemblage contains worked pieces of jet and ivory, as well as perforated snail shells and teeth.In 1998 excavations recovered a painted rock fragment, which might have been a painted piece of the cave wall, or alternatively, could have been mobile art (CONARD & UERPMANN 1999).
thin sections were observed with microfiche reader, transmitted polarizing light microscope, and electron microprobe analyzer (Jeol Superprobe JXA-8900R WD/ED Com bined Microanalyzer).Optical examination of the thin sections consisted of observation with the microfiche reader at magnifications of ca. 10 to 20x, and with the petrographic microscope at magnifications that ranged from 20x to 400x.Such a strategy allows the section to be viewed at different scales, ranging from field to hand sample down to ultra-fine detail with the SEM(COURTY et al. 1989).The large thin sections were also scanned on a flat-bed scanner(ARPIN et al. 2002) in order to facilitate documentation for the microprobe.This procedure enables efficient location of spots analyzed by the electron microprobe ana lyzer, and in turn correlate electron microscope data with those obtained through polarizing light microscopy.Polished thin sections of the undisturbed sediments were carbon-coated and analyzed by Jeol Superprobe, equipped with energy dispersive (ED)-and wave length (WD) dispersive-analytical techniques (see ScmrsGL et al. in press, for details).Finally, some of the loose samples of the burnt material from Layer 3cf (Figure 3) were analyzed by Fourier Transform Infrared Spectroscopy (FTIR; Nicolet Avatar 360 FT-IR E.S.E, com puter-controlled by Omnic Version 5.1).FTIR spectra within the wavenumber-range between 400 and 4000 cm"' were obtained by the potas sium bromide (KBr)-pellet technique.Tens of micrograms of sample were finely ground using an agate mortar and pestle, and ca.0.1 mg of the sample powder was mixed with about 80 mg KBr (Uvasol by Merck, IR-grade).A 7 mmpellet was produced by means of a hand press without evacuation.The empty sample chamber was used as reference background.Thirty-two scans with a spectral resolution of 4 cm"' spectra were carried out.If necessary, the baseline was corrected for measurements of the peak absorbances.Infrared-sensitive constituents were iden tified using a reference library containing more than 100 reference spectra relevant to archaeo logical samples.4ResultsMicromorphologyThe sediments from Hohle Fels Cave are more complex than they first appear on the macro scopic level in the field.The micromorphological examination of 17 thin sections from units C through A reveals fine variations of the texture, fabric and organization of the main constitu ents.Overall, the sediments are characterized by heterogeneity, which is itself expressed in different ways throughout the stratigraphic suc cession.The basic components, calcareous clay, lithic fragments, bone, charcoal and some fine organic material, are consistent through most of the sequence, although their organization and proportion change locally within the deposits.For the most part, the descriptions and discus sion below refer to the geological/archaeological subdivisions within C to A, the main units.The relevant subdivisions within Gravettian Unit C are layers 3c, 3b and their various subdivisions (Tables Fig. 5: Microphotograph of coarse, sand-sized clasts incorporated within rounded aggregates in sample HF-4B (Layer 3b).In particular, note the presence of charcoal fragments and rodent tooth in the center of the photograph.Plane-polarized light (PPL); width of view, ca.3.2 mm.Fig: 6: Microphotograph of sample HF-3C (Layer 3c) showing partially phosphatized limestone grain revealed by slightly lighter color on left and upper part of the grain (P with arrows; this is yellow-brown in color photograph).PPL; width of view, ca.3.2 mm.

Fig. 7a :
Fig. 7a: Microphotograph of phosphatized and etched calcite grain from sample HF-3C (Layer 3c).Etching of the calcite and replacement by phosphate creates an tooth-like pattern on the edge of the grain.PPL; width of view, ca.650 pm.

Fig. 12 :
Fig.12: Sample HF-7B (Layer lk).Microphotograph of massive subangular macroaggregates associated with fissures and crack structure.The matrix, bounded by the fissures, are constituted of cryoturbated, rounded aggregates compacted together with some looser matrix material.PPL; width of view, ca.3.2 mm.

Fig. 14 :
Fig.14: Sample HF-11 (Layer 3b).Microphotograph of loose rounded aggregates within an overall platy void structure.The platy structure, caused by ice lensing, is superimposed on a matrix that has already been subjected to cryoturbation.PPL; width of view, ca.3.2 mm.

Fig. 16 :
Fig. 16: Microphotograph of a thick, finely bedded silty capping in the central part of the photograph resting on a clast of oolitic limestone that makes up the lower half of the photograph; sample HF-3A (Layer 3b).PPL; width of view, ca.3.2 mm.

Fig
Fig. 19c: SEM BSE image of HF-7 with accompanying line scans shown in photo.Most of the scan is through the large, circular grain of calcite (cf.fig.19a, b), but also includes grains of quartz (high Si peak) and bone (peak of P and Ca).Iron is likely tied to the clay matrix.

Fabric
differences also serve to distinguish Lay ers 3c and 3b.The samples from Layer 3c are marked by loose, non-aggregated to loose aggre gated arrangements of matrix, often displaying platy voids (Figure 10).These voids are charac teristic of ice lensing in northern regions (VAN VLIET-LANOE 1982), and have been used as indicators of seasonal frost in cold paleoclimatic reconstructions (e.g., COURTY & VALLVERDU 2001; CREMASCHI & VAN VLIET-LANOE1990; Fig.20a: Microphotograph of the calcareous and calcitic matrix of the sterile layers, sample HF-1 (Layer Is).The sample consists of bright, rounded masses of calcite and calcareous ornaments within a calcareous matrix.Nore the cauliflower-type aggregate in the lower part of the photo, and the platy piece of bedded flowstone in the center right.PPL, Fig. 21: Sample HF-9B (Layer 3bt).Microphotograph showing the concentration of burnt and unburnt bones mixed within a calcareous matrix which commonly coats the bones.This is most likely accumulated as dumped burned material (see SOLVEIG et al. 2003 for details).PPL; width of view, ca.3.2 mm.
the choice of bone fuel over charcoal during the Gravettian occupation of the cave.Schiegi et al.(SCHIEGL et al. 2003) review the wealth of evi dence for bone burning during the Middle Pal aeolithic and Upper Palaeolithic of the Swabian Jura.On the contrary, during the Magdalenian occupation, there are fewer signs of use of bone fuel, but the presence of charcoal suggests use of regular wood as fuel.Finally, Layer 3bt within the Gravettian sedi ments (Figure4) was noteworthy in the sense that it contained a very high density of burnt bone.The micromorphological study demon strated the anthropogenic/biogenic nature of Layer 3bt within the Layer 3b sediments, which are impoverished in charcoal.At least two of these dark pockets of burnt bone and charcoal were seen in the profile (only one of which was sampled for analysis; Figure4).The burnt bone fragments range from slightly burnt to carbon ized and microscopic fragments are mixed with larger unburned bone pieces (Figure21).The few limestone clasts mixed with the densely packed burnt bone are unburnt and exhibit a capping similar to that observed in the Layer 3b sediments, suggesting that the formation of Layer 3bt took place during the Gravettian.PhosphatePhosphatization of the matrix or fine fraction is relatively high both in the Gravettian sediments (layers 3b and 3c) and the Magdalenian sedi ments (layers lk and Oc).The intervening sedi ments (layer 3a where present, is a mixed layer that separates lk from the Gravettian 3b layer), however, are markedly more phosphatized.Under ultraviolet light, the clay-rich matrix of Layer 3 c is homogeneously fluorescent, evi dence of its highly phosphatized and decalcified nature.Although Layer 3c includes roughly the same components -clay, silt, limestone, bone, and charcoal -as the layers above it, in Layer 3c these elements are finer and more "worked into" the matrix, either by physical or chemi cal processes.As displayed in sample HF-3A, Layer 3b contains a much coarser assortment of these components.This contact is visible at the macroscopic scale as well.In the field, 3b was described as much siltier and sandier -relatively devoid of clay -than the "argillaceous silt" of Layer 3c.The key contact between Layers 3c and 3b is also evident in the occurrence of large grains of phosphatic material in 3b in contrast to the relatively homogeneous alteration of the matrix found in Layer 3c.Ultraviolet examina tion of sample HF-11, taken from the lower part of Layer 3b, revealed dense phosphatic grains.These grains could be the remnants of a phosphate crust derived from guano or other products of animal activity.The presence of intensive animal occupation would account for the increased phosphatization of Layer 3c.The intensive human occupation and cryoturbation of 3b could have broken up this ctust, incorpo rating rounded fragments of dense phosphate within the overlying 3b sediments, as found in sample HF-11.The source of the phosphate is not totally clear.However, the abundance of bear remains from in the interior of the cave -which must have served as their hibernation home -would point to bear as a possible source of phosphate, either from the food they brought in or from their ex crements.There is very good evidence for the intense use of the cave by bears during Gravet tian times.Bones of cave bear are by far the most numerous, and the age profiles are dominated by young and old animals that usually die during hibernation (MONZEL et al. 2000).A number of mm to cm size pieces of carnivore coprolites was found in the thin sections.Although these are too small to identify as to specific animal, bears are a likely candidate.In any case, at this point the details of the geochemical and microenvironmental conditions responsible for decalcification and phosphatization of the ma trix, movement of phosphatic solutions, and the formation of phosphate reaction rims around limestone clasts are not clear and are currently under study.Conclusion This preliminary study of the Magdalenian and Gravettian sediments at Hohle Fels dem onstrates the potential of micromorphological study of archaeological deposits in cold regions, particularly this part of Germany which both underwent marked climatic change but also contains a rich record of prehistoric occupa tions.Although it is likely that no one meth odological approach could hope to completely elucidate the complex arrangement of anthro pogenic, biogenic, and geogenic elements de posited in this cave, micromorphological analysis provides information complementary to that gathered by traditional field and labora tory methods (COURTY et al. 1989).The Layer 3c sediments display platy structures indicative of more continuous cold conditions.While the overlying Layer 3b sediments also contain some platy or lens-like structures, they are dominated by rounded cryoturbation aggregates.These grains indicate a greater frequency of freeze/ thaw cycles, a conclusion supported by the presence of cappings of fine material on large clasts and aggregates.From the anthropogenic perspective, both Layers 3c and 3b are compat-ible with the use of bone as the ptimary fuel, while higher in the sequence, charcoal remains indicate the use of wood fuel instead (SCHIEGL et al. 2003).Much of this preliminary work will be refined as more exposures -both vertical and lateral -be come available at Hohle Fels.In addition, work at the contemporary site of Geissenklösterle about 2 km away will serve as a comparison to evaluate and extend the observations and inter pretations made here.This study adds further evidence to the discussion of human responses to climatic change in the Upper Paleolithic.When results from the Middle Palaeolithic and earliest Upper Palaeolithic units from Hohle Fels and Geissenklösterle become available, we will have some very clear notions about the environmental background at this important transition in human history.