“Rediscovery” of Paul Gervais' paleohistological collection

ABSTRACT Paul Gervais (1816–1879) developed an obvious interest for paleohistological researches during the latest phase of his career at the National Museum of Natural History, in Paris (Laboratory of Comparative Anatomy) and produced a collection of thin sections of extant and fossil eggshells, teeth and bone, although he did not published much on the subject. Current curatorial efforts have unveiled remains of this collection which is presented and assessed herewith.


INTRODUCTION
With the current development of histological researches in vertebrate paleontology, it is becoming more and more obvious that old collections of thin sections of fossil materials have a high historical interest and patrimonial value. Above all, they may be still useful for current researches. In spite of that, however, older collections of thin sections, and more generally of histological collections, are so far rarely the subject of careful curatorial eff orts in Museums, and especially in Universities (Ricqlès 2007).
Of course it is understandable that "routine" histological sections from easily obtainable laboratory animals are often considered as a consumable material which can be disposed of, or which can suff er high damage rates from the "attention" of students in the histological classrooms. However this policy cannot be accepted as soon as one is dealing with historical collections, extensive comparative collections from rare animal tissues, or collections which form the evidence of important experimental works. Histological collections from the past may not be easily duplicated. For instance, current -and welcome -nature conservancy regulations on rare or endangered species may hinder production of new extensive comparative histological collections, especially of growth series, and there are little incentives among modern biologists to start such kinds of researches anyway. Th is makes the preservation and curation of older histological collections of extant biological material a matter of special concern and even of urgency.
Th e situation is obviously even more stringent for the collections of thin sections from fossils because the general rarity or uniqueness of well preserved fossils enhances their patrimonial value. Whether they come from extant or fossil materials, in both cases the technical processes which have been used to obtain good histological sections have always been time consuming, labor-intensive and sometimes highly sophisticated. Only one large thin section may represent several hours of painstaking work for a highly skilled technician. Irrespective of any moral considerations, common sense suggests that it is a terrible waste of value, time, and energy for Science and Society at large to throw away such kind of collections, as it is too often the case when or after a researcher takes retirement, rather than preserve them for the community along sound curatorial practices, as demonstrated by the ultimate lucky fate of Professor Donald H. Enlow's collection (Ricqlès 2007).
Th is is all the more true now, than the current technical breakthroughs in image analysis and information processing and transmission and, by and large, generalized computerization, allow easy communication of data worldwide. It remains that histological collections should be now curated accordingly, forming "histothèques" or "histological data Banks" with access on line.
With this general framework in mind (Ricqlès 1987), we became involved in the research, inventory and preservation of older collections of thin sections of fossil vertebrates. Accordingly, some fi ndings on the fi rst thin sections processed from Saharian Dinosaurs (Ricqlès 1994) and from old sections from very early jawless vertebrates (Ricqlès 1995a) have been published. Similar works on the earliest thin sections from Pterosaurs have been performed (Steel 2003).
In connection with historical researches on the earliest studies of dinosaurs bones and eggs in Southern France, one of us (Ph. Taquet) noticed in the Collections of the Laboratory of Comparative Anatomy of the Muséum national d'Histoire naturelle, Paris (MNHN) an old collection of thin sections of fossil and extant birds and reptiles eggshells, performed as early as the 1870s at the request of Professor Paul Gervais (Taquet 2001). Th is collection apparently contains the earliest known thin sections of dinosaurs eggshels ever processed and is of lasting interest for researches on the early history of dinosaurs discovery in France (Taquet 2001). Further investigations in the Laboratory of Comparative Anatomy performed by Drs C. Cremières, M. Herbin, Fr. Poplin and one of us (V. de Buff renil) unveiled additional material in Paul Gervais' collections, namely a series of thin sections of fossil bones and teeth, which, as far as we know, were never described and fi gured in the scientifi c literature (Ricqlès 2006). Paul Gervais (1816-1879), a well-known comparative anatomist and palaeontologist (Laurent 1996) who published extensively on fossil mammals, remained doubtful on the general signifi cance of evolution. Gervais' own use of his collection of thin sections appears to have been limited to a two part publication (Gervais 1875a, b).
Th e purpose of the present work is to off er a listing, brief description, selected illustrations and assesment of this collection. As it mostly deals with fossil mammals, we felt advisable to dedicate this survey to our good colleague and friend, Professor Louis de Bonis, who himself contributed to some paleohistological endeavors (Bonis et al. 1972).

MATERIAL AND METHODS
As currently preserved, Paul Gervais' collection of fossil thin sections consists of several boxes. It is obvious that the collections experienced numer-ous circumstances over the decades, involving box changes, successive cataloging and labeling eff orts, and discarding of sections in poor condition, to make room for newer preparations (Crémière 2004.). Th e boxes were numbered and given general titles such as "oeufs" (eggshell thin sections), "os, cartilage" (bone and cartilage thin sections) but no attempt was apparently ever made to catalog each thin section individually. One small box contains 45 slides (ground sections on standard 7.5 × 2.6 cm glass plates with round coverslip) of eggshells from extinct and extant birds and reptiles (Taquet 2001). A larger box contains almost 100 slides (thin sections) of fossil bones and teeth, similarly mounted. Only the later, still undescribed up to now, will be dealed with here. With few exceptions, the technical quality and general preservation of the sections have been found to be rather satisfactory. We stress that the collection described here may be only a part of Paul Gervais' original collections of fossil thin sections possibly still surviving in the Laboratory of Comparative Anatomy of the MNHN: some isolated thin sections apparently from the same collection have been found scattered in other boxes and further inquiries are in progress. Th e additional fossil thin sections found elsewhere but presumably part of the same initial collection, according to Drs C. Crémière and M. Herbin, have been also described here (they are noted N-1 to 13). Moreover, a great number of comparative thin sections of dry bone and scales from various extant vertebrate species are also preserved. Th ey are not dealed with here and considerable further curatorial and descriptive works obviously lie ahead. Indeed several other old histological collections, including bone thin sections, were performed at the Laboratoire d'Anatomie comparée during the late 19th century, at considerable technical cost, although they have been apparently of little scientifi c use so far, for various reasons (Crémière 2004;Ricqlès 2006).
Each section has been fi rst carefully cleaned from dust and examined regarding the various old numbers or labels that it may bear (Fig. 1A, B). It was then observed under a coumpound microscope in ordinary and polarized transmitted light. Th e original labels read (as far as possible) on each section are reported below ne varietur. Uncertain- Ricqlès A. de et al.  ties are quoted between parentheses with question marks, and further discussed (see Comments). We found unwise to add any new label or number to the sections. Accordingly, the short descriptions below refer each section by the original labels and number(s) it bears. Irrespective of their order as found in the boxes (or elsewhere: sections "N-1 to 13"), they are described herewith seriatim, following a numerical sequence roughly refl ecting current taxonomic order. Description. Th e two slides may correspond to very thin scale-like materials observed in toto rather than to actual thin sections. In both cases, the two objects, of a few square milimeters, have an irregular shape with broken edges. Th ey are formed by a homogeneous brownish material, monorefringent under crossed nicols. Th e whole surface is marked by very numerous thin undulating crests, roughly parallel to each others and, likely, also to the external shape of the structures (when preserved). Some crests locally bifurcate. Some faint higher order cycles (4-5) might be deciphered; each cycle containing about 12-15 crests. Th e crests are strongly reminiscent of the circuli in the teleosts elasmoid scales. Similarly, some radial thin clefts suggest scale radii, although they may be artefacts. Very thin, numerous dark radial or oblique fi bre-like structures are locally observed, although they may also be artefacts, at least in part. Th ey could express the plywood-like organisation of the collagenous fi ber bundles forming the underlying basal plate of the scale. No vasular or cell spaces could be deciphered.

DESCRIPTIONS
Comments. Stensiö (1969: 646-659) described macroscopically and commented the scales in Coccosteus Agassiz, 1844-1845 and other Placoderms. He noted the limited histological knowledge of this material at that time and interpreted the condition in Coccosteus scales as a secondary simplifi cation by reduction of the deeper components of the more typical three layered dermal skeleton (Stensiö 1969: 652-653). Whatever it may be, the structures strongly agree in general aspect with teleosts elasmoid scales of the cycloid subtype (e.g., Meunier 1981; Sire 1986). Description. A longitudinal section through an elongate, thin tooth-like organ, about 12 mm long and 3 mm wide at its base. Th e section shows two diff erent tissue types, a and b.
Tissue a: all the periphery is formed by a homogeneous tissue of constant thickness, birefringent under crossed nicols. Th is tissue is itself divided in three regions. a1: the most external one is characterized by very numerous thin dark canaliculi perpendicular to the free surface. a2: the middle one is almost devoid of canaliculi and shows a faint organisation of the tissue as very thin "sheets" layered parallel to the free surface. a3: the inner one, with a complement of canaliculi similar to a1 but far less numerous. Th is lesser density allows observing some branching patterns of the canaliculi, several branches uniting from the periphery towards the center.
Tissue b: the internal part of the organ is made of a cone which almost reaches the surface at the tip of the tooth. Th is cone is formed by a highly birefringent vascularized tissue. Numerous vascular canals of small diameter are lining the limit between tissues a and b. Large and small vascular canals with irregular branching patterns are spread deeper within tissue b. Actually, there is no structural discontinuity at the limit between tissues a and b, but only a change in the orientation of canaliculi and tissue fi bers, plus vascularity restricted Ricqlès A. de et al. to tissue b. Th e canaliculi are numerous in tissue b, but highly irregular, forming locally a very dense meshwork. Th ey are far less numerous or almost lacking in the material immediately surrounding the vascular canals. Th ere is nothing like a pulp cavity in the structure.
Comments. Th is thin section is very interesting from an historical-technological point of view. It is still technically excellent. Th e glass plate is not standard but very small (54 × 17mm). Th ere is no cover slip and the section itself is only protected by a blue paper glued on both faces of the glass plate, with a circular opening around the section itself. Th is paper also bears the original label in English. Very likely the section was processed in Great Britain sometimes during the early development of the thin section technique.
Histologically, we interpret "tissue a" as a kind of orthodentine, forming the so-called "mantel dentin" at the periphery of a tooth. "Tissue b" would be a so-called "circumpulpar dentin", actually a vasodentin, with centripetal deposition of dentin around vascular canals, the denteons of Ørvig (e.g., 1967), analogous to the primary osteons of bone. Th ere is no evidence of any enameloid.
Description. Two partial cross sections of simple labyrinthodont teeth, still partially embeded in the matrix. Th e pulp cavity is free, with about 12 short radial expansions of dentine protruding in it. Each inward expansion roughly coincides with a slight groove at the external surface of tooth. Tooth material is monorefringent under crossed nicols. Th ere is no distinct coating of enamel. A thin region of globular dentin forms the tooth material at a small distance from its external surface. Inside, the tooth is formed of orthodentin with very long, thin, strait canaliculi within and between the radial expansions. Overall, the material agrees well with Schultze's (1969) illustrations of Lyrocephaliscus Wiman, 1910 (Schultze 1969: pl. 21 fi g. 3, pl. 22 fi g. 2), a Trematosaur. 5. "Archegosaurus latirostris de Lebach 1870-480". Pencil number: 1073, ink label: Anat. Comp. 1875 (Fig. 1D).
Description. One isolated tooth free of matrix. Th is cross section may come from a region somehow closer to the tooth tip. It shows about 20 radial expansions into the free pulp cavity. Long expansions tend to alternate with short ones, and the long ones slightly undulate. Again, no enamel could be deciphered. However the section slightly shows the radial expansions of the external dentine forming the core of the radial fences. Description. About 10 osseous fragments are scattered under the coverslip. In most fragments the bone tissue is sectioned longitudinally. Osteocyte lacunae, more or less plump or elongate in shape, are set in line with the parallel fi bered bone tissue. Some vascular canals, forming primary and secondary osteons are also sectioned longitudinally. In two fragments, the structure is more complex, with several vascular canals parallel to each others or faintly converging in a fan-shaped arrangement. Sharpey's fi bers are numerous after a faint annulus. Th is structure is suggestive of the bone not far away from a tooth insertion. Description. A few square millimetres probably from a horizontal (oblique) section in the external cortex of a dermal bone. Th e light brown bony material is mostly birefringent under crossed nicols, especially in regions around vascular canals (primary osteons). No secondary osteons (Haversian systems) are observed. Osteocytic lacunae are diffi cult to observe. Noticeable locally are Sharpey's fi bers and superposition of undulating dark lines strongly suggesting annuli or LAGs (= lines of arrested growth), as well known among Stegocephalians (e.g., Ricqlès 1995b).
Description. Probably a horizontal section in the external cortex of a dermal bone, broken in several small fragments, and further damaged by balsam shrinkage. Bone is strongly birefringent under crossed nicols, apparently demonstrating two orthogonal superimposed plies of bony lamellae. 9. "Labyrinthodon de Stuttgart. Écaille? temporale". Pencil number: 1077, ink label: 1938-613. Charles Marchand, Préparateur à Paris. (Fig. 1F) Description. A good section from the free surface of a bone (with matrix) to about 3 mm deep in the cortex. No superfi cial ornamentation is obvious. Th e bone is generally birefringent under crossed nicols with a clear lamellar organisation. It contains about 16 superimposed rows of vascular canals of small diameter, sectioned longitudinally and circularly, and organized as primary osteons laid down in a "lamellar-zonal" matrix of periosteal origin. However, neither Sharpey's fi bers, nor cyclical annuli could be observed. Secondary osteons seem to be lacking. Description. A poorly preserved bone fragment, probably sectioned vertically. What might have been the external, ornamented surface appears to have been mostly replaced by fan shaped, radiating black mineral structures which could be superfi cially mistaken for Sharpey's fi ber bundles. Actually, very little of the external bone is preserved. Most of the specimen is formed by thin, remodelled endosteal bone trabeculae forming a highly cancellous tissue. In many places, osteocytic lacunae appear to have experienced localized widening, probably the result of post mortem destruction. Indeed, evidence of boring canals of mycelial origin can be observed locally.
Comments. Occurrence of Trimerorachis in the collection and the presence of Professor Edward D. Cope's name on the section label demonstrate the relationships among French and American paleontologists of the era. Description. A section containing several "Prototriton" (sic) bones still in their relative positions in the matrix. What can be observed suggests a complete small skeleton sectioned frontally, e.g., parallel to the bed plate in which it lies dorso-ventrally. Unfortunately the technical quality of the section is poor. It nevertheless allows deciphering occurrence of several bones and teeth. At least eight teeth can be observed in situ on teeth-bearing bones. Th e teeth are non labyrinthodont in structure, with a very thin dentin surrounding a relatively large pulp cavity. Th ey appear to be pedicellate as among most Lissamphibians and a region of intensive remodelling, where a tooth base ankyloses on the dentigerous bone is visible. On the other hand neither enamel or enameloid, nor bicuspidate structure could be observed. Various parts of the skeleton, including dermal bones, and short and long endoskeletal bones, are sectioned more or less at random. Bone tissue shows little details but appears to have been poorly vascularized. Some long bones sectioned longitudinally may show sequence of endochondral ossifi cation in the calcifi ed cartilage of metaphyses. However the poor quality of the section and randomness of orientations preclude detailed observations.
Comments. Th e cover slip of this section is not round (as on most sections in the collection) but rectangular, as in modern thin sections. Th e endoskeletal bone structures preserved in this specimen may be compared to those of post metamorphic Urodeles (Ricqlès 1965).

"Prototriton de Gaudry. Permien d'Autun". No numbers.
Description. A large thin section, obviously from the same material as above (11). Preservation is very poor (see Comments). Th e section is rather thick but allows deciphering a longitudinal section in a limb with the stylo-zeugopodials in articulation (no autopodial is preserved).
Comments. Th is section has a large oval-shaped coverslip. Th is has largely separated from the glass plate over time and gently curled. Th e balsam is badly cracked and the section will have to be intensively cured for further preservation and observation. Description. A small (4 × 4 mm) fragment of cancellous bone. Th is spongy bone tissue is formed by rather coarse irregular trabeculae, leaving between them irregularly elongate spaces. Th e bone tissue has a great cellular density and shows extensive erosion/reconstruction cycles, as well as osteon-like structures within the thick trabeculae. Some aspects within the trabeculae are suggestive of globuli ossei. If so this tissue would originate in part from previous endochondral ossifi cation.
Comments. Another thin section of English origin, exactly similar to section 14 (N-3) (see below). It is not possible to decipher the exact label name; both "Plathyosaurus" and "Plathioposaurus" can be read. Th e former name does not appear to match any genus name coined to a fossil Amphibian or Reptile during the 19th century. Th e later name appears close to Platyposaurus, which would be a synonym given by Lydekker, 1890 to Platyops Twelvetrees, 1880, a Permian "Rachitome" (Lehman 1955).

Description.
A bone tissue fragment of 5 × 5 mm. Th is is a microcancellous bone tissue spatially organized along a very regular "honeycomb" pattern. Th e thickness of the section prevents from observing details of the tissue itself.
Comments. One of several sections probably pertaining to Paul Gervais' collection but of English origin. Th e section is set on a standard (7.6 × 2.5 cm) glass plate with a round cover slip. Th e glass is entirely covered (and protected) by a red paper below, a red and gold paper above; with the label "C. M. TOP-PING" appearing twice into circles on top.

Description. A good cross section of a small bone.
Th e section is irregularly oval, about 5 mm in its largest diameter, and with one fl at, slightly concave face. Th e cortex merges very gently with the more cancellous central region, as usual among aquatic tetrapods (Ricqlès & Buff rénil 2001). Th e outer cortex is primary in structure, containing longitudinally oriented primary osteons embedded in a woven tissue, radially oriented in the thickest par of the cortex. In the thiner (more slowly growing) regions, the cortex shows less vascularity, no radial orientation and some evidence of cyclical deposition. Th e medullar region is rather dense, small marrow cavities being set apart by thick, irregular secondary trabeculae of endosteal bone tissue. No remnants of cartilage are observed.
Comment. Th e thick convoluted trabeculae in the medullary region agree with the (non pathological) "osteosclerotic" condition often observed among aquatic tetrapods (Ricqlès & Buff rénil 2001).
Description. Only cancellous bone tissue can be observed in cross section. Th is is a "mature" bone tissue with evidence of erosion/reconstruction cycles forming areolar secondary endosteal trabeculae. Some structures agree with the large endosteal secondary osteons lining the marrow sinuses often observed at the periphery of the medullar region in many large tetrapods.
Comments. Pollution by grinding powder hinders detailled observations. Th e structures are rather similar to the ones observed in an adult Plesiosaur phalanx (see Ricqlès & Buff rénil 2001: pl. 4, fi g. F.).

Description.
A large portion of cancellous bone retaining its microanatomical organisation, possibly suggesting longitudinal orientation of bone trabeculae along the section, in a vertebral centrum. Otherwise the section is too thick to show histological details.

Description.
A partial cross section of a bone of at least 10 mm in diameter. A central free marrow cavity (diameter 4 mm) contains a few bone trabeculae. Th e remains of the cortex is 5 mm thick and entirely formed by a very cancellous tissue, somewhat organized radially and longitudinally. No histological details can be deciphered because of bone opacity.

Description.
A partial cross section from a large tooth (partial diameter 14 mm), close to its base. Th e tooth has a complex, "folded" structure, showing 9-10 partly preserved folds of dentine circling a large pulp cavity entirely fi lled by fi nely cancellous bony tissue. Each individual fold is a U-shaped mass of orthodentine of about one millimetre thick, the top of the U opens towards a small pulp cavity. Th e orthodentine has typical dentinal tubules, oriented from the periphery to the pulp cavity along radial curves, crossed by several (6 to 8) "growth lines" parallel to the external surface of the folds. Close to the surface, a thin coating of very dark dentine appears to be globular and is capped by a very thin sheet of a strongly birefringent hard tissue with a slightly irregular surface. Between the folds a distinct bone-like tissue contains cell spaces and vascular canals, apparently in structural continuity with the highly birefringent thin sheet mentioned above. Th is osseous tissue (topologically located outside the dentine) is in continuity with the fi nely trabecular bone fi lling the pulp cavity. Th e dentine folds are isolated from each others at the level of this cross section.
Comments. Th is section is of a rather good quality. It is mounted on a tiny glass plate, with original label on blue paper, as section no. 3 "Lamna" fossil (see above) but better prepared. Th ere is no coverslip and the section is thus technically very similar to section no. 3 and may have the same British origin. Th e tissue between the dentine folds may be interpreted either as a cellular cementum (as suggested by its specialized fi bre systems) or as bone of attachment, nevertheless it merges with the regular bone inside the pulp cavity. Th ere is no clear indication of enamel.
Th ese structures agree with later descriptions of ichthyosaurian teeth by Schultze (1969: 111-115, text fi g. 18 and pl. XVI fi gs 1, 2) and Schmidt & Kiel (1971). Description. Bone tissue is observed on a 20 × 5 mm surface. Th is section is of fairly good technical quality and allows detailed observations. Th e preserved free surface of the bone is fl at and formed by a very thin cortex which likely experienced subperiosteal resorption. Th is cortex has the same histological structure as the inner trabeculae which merge with it locally. Inside the bone, the tissue is cancellous, formed of very irregular trabeculae. Th ese trabeculae are of endochondral origin, as evidenced by the presence in their core of many island of calcifi ed cartilage, with their clusters of large chondrocytes lacunae. Th e globular mineralisation of the cartilage matrix is also conspicuous, as well as globuli ossei localized at the interface between bone and cartilage (cf. Buff rénil et al.1990).
Comments. Th e resorbing trabeculae forming the cortex retain an endochondral component. Th is structure is comparable to the surface of long bone metaphyseal regions which have to reduce their outer diameter (modeling) during the growth in length of the bones, so that the overall morphology of the bones can be maintained while metaphyseal regions are sequentially relocated (Enlow 1963).

Description.
A cross section of a rather compact bone fragment (jaw), about 17 mm long and 6 mm maximal width, in part contained in matrix. Th e preservation is fair, although the balsam has badly shrunk and cracked below the cover plate. Th e subperiosteal bone tissue shows incorporation of new vascular canals, Sharpey's fi bers and "growth cycles". Th e osseous tissue shows evidence of post mortem degradation, suggesting that the bone was stranded on shore before ultimate fossilization. Th e inner bone is extensively converted into Haversian systems and contains large "marrow" sinuses, one tooth socket and some tooth material. However it has entirely experienced diagenesis and no dental tissue structure remains.

Comments.
A striking artifact is provoked locally by the schrinken balsam, that displays appearance of a globular tooth with an "onion sheets" structure under low magnifi cation. Comments. Superfi cial (sub periosteal) bone layers do not seem to be preserved in the section. Description. A 8 × 8 mm cross section in coarse cancellous bone. Th e section is rather thick but the bone structure can be deciphered. Th e bone trabeculae are irregular and thick, with small marrow spaces between them. Th e bone tissue is secondary and remodeled. It appears mostly endosteal in origin.

(N-6) -
Comments. One of the early thin sections of British origin in P. Gervais' collection, and one of three with the same, careful technical setting under red and gold paper. Th e material should come from an Iguanodon Mantell, 1825 fragment from England antedating the Bernissart discovery; it is so impossible that this specimen comes from the Belgium material. However, the origin of this specimen is not precisely known (England, Isle of White?).

Description.
A cross section probably in the external cortex from the shaft of a long bone. Bone tissue is entirely primary in structure. Th e section is polluted by abrasive grains but the structures can nevertheless be observed. Th e bone tissue with a dense, homogeneous and regular vascularisation organized as primary osteons, is typically laminar to sub-plexiform. Th e bone surface, as preserved, shows Ricqlès A. de et al. no evidence of an external fundamental system or a decrease in appositional rate. However it might not represent the natural subperiosteal surface of the bone. No lines of arrested growth, or Haversian substitution are clearly visible.
Comments. Th e structure suggests an immature, actively growing individual.
Description. An excellent thin section in compact bone, about 10 × 9 mm, plus some additional fragments. Th e main part of the section is a cross section through a massive cortical bone entirely formed by primary (periosteal) bone tissue of laminar to sub-plexiform types. Only a few vascular canals have experienced Haversian reconstruction. Five small additional fragments under the same cover slip fi gure the same tissue, but cut tangentially, parallel to the bone surface. Th ey show three complex (and diffi cult to interpret) interwoven components: 1) the fi brous component of periosteal origin; 2) the lamellar component of the primary osteons (endosteal component); and 3) the vascular component locally appearing as a reticulum (e.g., Ricqlès 1972: fi gs 6, 7).
Comments. Th e structure, similar to section 24 (see above), should originate from the same material. It also suggests a fast, massive, almost continuous new bone apposition in a presumably juvenile to half grown individual. Th is thin section, again, raises the issue of the origin of the material. It looks to have been processed in Paris, along most others from Gervais' collection, possibly by E. Bourgogne or C. Marchand (although no label suggests this). If so, and if the material indeed belongs to Iguanodon, it should have come from a fragment sent from England. An alternative is that the material comes from Southern France (see below) and was labelled Iguanodon by comparison with other sections, published or not.
Description. Two small irregular fragments under a square coverslip. Th e bone tissue is mostly cancellous, and only one small region seems to agree with a bone cortex of primary (periosteal) origin. Th is cortex is densely vascularized by up to fi ve rows of primary osteons organized into a vaguely laminar pattern. Th e spongiosa starts abruptly in the deep cortex, which is dissected by large marrow cavities. Th e endosteal trabeculae are very remodelled. Th e whole bone experienced extensive post mortem erosion by boring organisms.
Description. A small bone fragment, about 7 × 5 mm in longitudinal section. Th e tissue is highly birefringent throughout. A small region may correspond to the deep primary cortex, with a few Sharpey's fi bers and rather abundant anastomosing vascular canals sectioned longitudinally. However, most of the section is secondary endosteal and would form the external part of a rather dense spongiosa.
Comment. Again, this section does not bear much "diagnostic" characters but the structure could generally agree with a dinosaurian origin. Comments. Th e section suggests an immature, fast growing bone, perhaps exposed to peculiar biomechanical demands for a land dwelling mammal. Description. A small longitudinal section (4 × 10 mm in two parts) in compact bone. Th e natural subperiosteal surface of the bone is exposed on one side of the section, while the endosteal (perimedullar) region is exposed on the other side. What can be deciphered is consistent with the preceding section.
Description. A small longitudinal section (7 × 9 mm) in compact bone. In this better preserved section, the primary structure of the densely vascularized bone cortex is easily observed. Tissue (a) is monorefringent. It is crossed by very numerous fi ne, non undulating canals, or very elongate cell spaces, with a distinctly observable diameter, running parallel to each others and with indications of short branching of oblique lateral expansions. On one side, tissue (a) merges in contiguity with tissue (b) by a very thin region where tissue (a) canals almost disappear. On the other side, tissue (a) is set apart from tissue (b) by a very regular, non undulating "cementing line". Tissue (b) is cellular and vascular. Cell spaces are numerous, plump and slightly polygonal; they are embedded in a birefringent matrix showing two orthogonal fi brillar directions. Th e matrix is also permeated by a large number of "canals". Th ese are mostly parallel to each others and run towards tissue (a). Close to the cementing line with tissue (a), they change their direction and tend to run parallel to the cementing line. Around each "canal", the tissue matrix is strongly birefringent, demonstrating a primary osteon-like structure with two successive orthogonal directions of fi bers. Tissue (c), sometimes observable at the periphery of tissue (b) is monorefringent and badly preserved. Many sand grains permeate the "tissue" which otherwise would show remnants of a dark cloudy matrix with some vascular canals locally. 36. "Megatherium No. 23" (some other words are undecipherable and had been crossed; Fig. 2E).

Description.
A very small section, prepared with blue paper as section no. 3 "Lamna fossil" (see above). It shows about 4 × 3 mm of tooth material. Description would match tissues a and b above but orientation in the structure diff ers; moreover, preservation and optical qualities are better. Th e "canals" in tissue B are here crossed longitudinally and show some branchings. Canaliculi in tissues a and b meet at an angle at the level of the cementing line between them.
Comments. Since Owen's time (1840-45), the peculiar tooth histology of "edentates" has been a subject of study. According to traditional views (Schmidt & Kiel 1971: 148-152, fi gs 104, 105), tissue (a) would match the hard or "compact" dentin forming the tooth crests; tissue (b) would be the so-called vascular dentin while tissue (c) could represent remains of cementum.  (Fig. 2F).

Description.
A cross section in compact bone, of a roughly triangular shape, with 9 mm of bone surface extending 6 mm in bone depth. In this good thin section, the primary cortex is formed by a grossly lamellar tissue of periosteal origin. Bone fi bers alternate under crossed nicols. Vascularisation is provided by small primary osteons organized longitudinally. Th e canals are arranged in about fi ve, poorly defi ned concentric rows. Short radial vascular anastomoses connect some canals. A dense Haversian tissue with typical, well-lamellated secondary osteons, spreads by places in the primary cortex up to its surface. Th e same tissue also forms the remaining deep cortex. At the endosteal margin, the deep cortex is limited by large perimedullar cavities lined by thick endosteal trabeculae.
Comments. Th e primary cortex suggests radial growth at very moderate rate, moreover, extensive Haversian substitution does not suggest a juvenile condition. However, the lack of a typical "external fundamental system" and the relatively small amount of superimposed generations of secondary osteons both suggest a less than fully mature condition. Amprino & Godina (1947: 351-52, 452, fi gs 64, 66) describe the histology of Bradypus Linnaeus, 1758 femur and tibia as densely Haversian with locally slow growing primary bone at the periphery. Th is does not allow substantiating histologically the taxonomic status of Filhol's "Bradypoid". Description. A small section (5 × 3 mm plus detached fragment) in rather compact bone. Th e section is similar to the preceding one, but the structures are cut more transversely. Secondary osteons appear well delineated, in cross section, from neighbouring tissues. Th e tissue may be "compacted coarse cancellous bone" (Enlow 1963)

Description.
A large, irregular section (15 × 20 mm) of a dense, broken bony material (and perhaps teeth). As in preceding sections, the well preserved tissue is highly birefringent, with most fi bers of a given osteon apparently organized longitudinally, as are also the osteocyte lacunae. Th e cell canaliculi are mostly oriented perpendicular to the fi bres. All the bone appears as built of secondary osteons, forming very distinct "pipes" in an overall compact bone tissue, both in longitudinal and cross sections. Th is aspect refl ects the presence, in the thin peripheral layer of each osteon, of fi bres orthogonal to the longitudinally-organized fi bres forming most of the osteons. No tissue in the section can be interpreted as dentine or cement. Description. Two small fragments (2 × 2 mm) in tooth material in cross section. Th e thick wall of a hollow cylinder is apparently preserved. Action of boring organisms is evident. Th e tissue is slightly birefringent. It contains numerous cell lacunae, no Ricqlès A. de et al. vascular canal and is entirely permeated by long, thin fi bres oriented radially.

Lestodon
Comments. Occurrence of numerous autochthonous cell lacunae identifi es the tissue as cementum rather than dentin.
Description. An elongate (22 × 5 mm) longitudinal section in compact bone. Th e putative natural surface of the cortex is badly preserved because of post mortem boring organisms, which precludes histological analysis. Deeper, the better preserved tissue is apparently of the dense Haversian type, with secondary osteons cut longitudinally or transversely. Th is tissue forms a dense but thin cortex. Still deeper, a well developed endosteal margin is formed by large perimedullar cavities lined by secondary fi nely-fi bered endosteal bone tissue.

Description.
A longitudinal section (9 × 5 mm) in compact bone. Th e tissue is slightly birefringent. A very thin primary cortex contains a few primary vascular canals and some Sharpey's fi bers. Th is cortex is immediately replaced below the surface by Haversian tissue forming the rest of the cortex thickness. Scalopped cementing lines at the periphery of the secondary osteons are well marked. Th e osteocytic lacunae included in osteonal lamellae are not all organized longitudinally. Instead, their orientations appear to alternate from lamella to lamella within and between osteons. No endosteal margin is preserved.

Description.
A slightly larger (10 × 7 mm) longitudinal section in compact bone. No remain from the superfi cial cortex is preserved. Otherwise bone structure is similar to that of the preceding section, which can be assumed as coming from the same bone. Description. A rather small (9 × 5 mm) cross section in compact tissue. Structures are almost entirely obliterated by the poor tissue preservation. Only a thin, grossly lamellated external cortex with few longitudinal vascular canals, and possibly radial Sharpey's fi bers, may be deciphered. Description. Th ree fragments of rib in cross section. Apart from a putative, small and very thin remain of the external bone surface, all the tissue is dense Haversian. Th e tissue is largely monorefringent with exception for the most external region of the secondary osteons that shows high birefringence. Several generations of osteons are observed, as well as fully open erosion rooms, suggesting a still active erosion/reconstruction process.
Description. Two fragments in cross section. Th is dense Haversian bone gradually merges laterally into a more concellous (inner) tissue formed of strong endosteal trabeculae. Th e Haversian tissue is more anisotropic than that of the rib, suggesting a diff erent spatial organisation of its (former) collagenous fi bers. Th e free bone surface, as preserved locally, is formed of a very thin primary tissue, rich in Sharpey's fi bers, broadly replaced below the surface by Haversian substitution.

Description. A larger (12 × 13 mm plus isolated fragments) longitudinal section in compact bone.
Th e tissue matches the dense Haversian structure observed in cross section. A few oblique to transversal osteon orientations and the pattern of anisotropy of the tissue suggest a more complex spatial organisation of bone fi bers than that in the rib.

Description.
A slightly larger (9 × 10 mm) longitudinal (tangential) section in well vascularized compact bone. Th e tissue is highly birefringent and is formed of secondary osteons and endosteal deposits rather regularly organized. Many secondary osteons appear to be composed of parallel-fi bered rather than fi nely lamellar tissue. Descriptions. Th ree cross sections in compact cortical bone tissue. Similar circumstances caused slightly less devastating results for observation in the cross sections than in the longitudinal ones. From the periphery, the cortex varies in thickness from 5 to 10 mm. A very thin lamellar external fundamental system appears; otherwise the cortex is densely Haversian throughout. Th e periphery of each secondary osteon tends to be better preserved than the core of the osteon. G. Marchand, Préparateur. (no. 67 is illustrated on Fig. 3B).
Descriptions. Eleven thin sections (one broken) in osteoderms. Each section is sub-circular, varying in diameter from 5 to 10 mm. Th e material is compact, with no central cavities forming a spongy region. Th e periphery of the osteoderm is formed by an extremely dense, compact tissue, almost non vascular, with radial fi bres, perpendicular to the surface. Th e tissue is organized into numerous (> 12) successive thin strata, parallel to the surface. Th is organization appears to result from the alternation of thinner, translucent, and thicker more opaque strata, that diff er in their general orientation, rather than in the nature of their fi brillar material (as also suggested by reactions under crossed nicols). In some cases this stratifi cation results into a more clear-cut alternation of opaque zones and translucent very thin lines; the zones becoming thinner and thinner towards bone surface. Cell spaces are numerous but not conspicuous. Th is organization may be followed, more or less unchanged, down into the core of some osteoderms. In others, the general system of radial fi bers becomes more conspicuous in the deeper regions, forming very elongate, sometimes fan shaped systems of compact fi bers bundles. Th ose systems are set apart and lined by others, orthogonal to them, and forming together a complex 3D dense meshwork. A faint system of vasular canals occurs in the osteoderms, especially in the deep regions that were formed during early osteoderm development. Th e canals are irregularly oriented, with some short anastomoses between them. Th ey appear to be either of the primary simple vascular canals type, or to develop as primary osteons. Haversian reconstruction, if any, seems to be exceedingly rare. Some post mortem cracks have been fi lled by a dark mineral (Manganese oxide?) which locally forms "ferns leafs-like" images. Locally, this material has naturally stained the structures in a faithful way, either from the surface or from the inner cracks, enhancing the optical contrast between the neighbouring fi brillar systems.
Comments. Th e osteoderms appear to be formed by a rather specialized type of dense primary bony tissue, also observed in some dinosaurs osteoderms , where an extensive system of closely packed fi bers takes prevalence over a weak system of vascularisation. Th e core of the osteoderm may have been initially formed here by the metaplastic ossifi cation of the preexisting fi bers of the stratum compactum of the dermis. Subsequent growth may have taken place by further deposition of new material at the periphery of the core, although there is a clear fi brillar continuity between deep and more superfi cial regions. Growth cycles at the periphery of the osteoderms strongly resemble the annuli and zones described in the bones of many extant ectotherms (Castanet et al. 1993), where they have been demonstrated to be annual in many cases. Th e whole comparative analysis of osteoderms formation and growth among various tetrapod groups still remains a little surveyed problem.

1878.
Description. A small (9 × 5 mm) longitudinal section in compact bone. Most of the fi ne structure has been destroyed by post mortem boring organisms. A very thin but well-vascularized primary cortex, with vascular canals running in several directions, is replaced at a small distance from the bone surface by longitudinally organized Haversian tissue.

Description.
A large tooth thin section with a rectangular coverslip. Th e tooth is broken in several fragments by the technical process but tissue structures can be observed. Most of the material, cone shaped, appears to be orthodentine, although preservation of the tissue is very poor, due to post mortem invasions. Th is cone appears to have been covered by a thin (about 400 μm) continuous sheet of a highly birefringent tissue which remained in situ, at least locally (Fig.  3C), and did not suff ered invasion by boring Ricqlès A. de et al. organisms. Th e tissue shows a very fi ne radial striation throughout and faint suggestion of an irregular globular structuring of its matrix. No obvious "prisms" could be recognized. On its external surface (opposed to what we interpret as dentin) this tissue may have an irregular surface, with some undulations, bulgings and shallow grooves locally. In some tooth fragments, this tissue displays, in transmitted polarized light, at least 17 to 20 regularly spaced incremental layers (interval: 25 to 30 μm) that are sharply defi ned in the strata close to the undulated surface, but tend to vanish towards the opposite surface (Fig. 3D).
Th is tissue is interpreted as an enamel of some sort, and the striation that it displays consists in striae of Retzius. Externally, a third tissue, very poorly preserved, as is dentin, shows cell spaces, short fi bre-like striations, and some local evidence of cementing lines. It is interpreted as cementum.
Th e dental region shown on Fig. 3C is of special interest because it shows the end of enamel as an edge at the crown-collar junction, after which the cementum is directly capping the dentin, with only a line of discontinuity between them. Description. A small tooth under a circular coverslip and dark mounting medium. Th e tissue preservation is correct but tooth exploded during the technical process, with several fragments becoming superimposed. Obliquity of the section in the enamel and cementum provides further details: the enamel shows cycles of diff use, cloudy transverse "bands" in polarized light, but no obvious prisms. In the cementum, plump cementocytes lacunae and bundles of long fi bers are associated with numerous simple vasular canals of small diameter, running parallel to each others. Th e cementum is thus vascular. Its location in connection with enamel suggests it is crown rather than root cementum.
Description. A small triangular (side 7 mm) thin section in spongy bone. As already recognized on the label, the intensity of post mortem destruction by boring organisms makes histological study diffi cult, although the overall microanatomical organization of bone trabeculae (and bone porosity) can be still easily assessed under crossed nicols.
Description. A small quadrangular (side 7 mm) thin section in compact bone. Same comments as above.

Description.
A quadrangular (side 11 mm) longitudinal thin section in compact bone. Th is section is badly, but not completely, invaded by post mortem boring organisms. Th e remain of the bone tissue suggests a longitudinal section in a dense Haversian tissue, with prevalence of longitudinal orientation of bone fi bers. Apparent active biological resorption of the bone may rather represent the deceptive eff ect of boring organisms.

Description.
A large quadrangular (broken in two, sides 11 × 7 mm) longitudinal thin section in compact bone. Same comments as above.

Description.
A large irregular oval (16 × 9 mm) cross section in compact bone. Parts of the free surface of the external cortex are preserved. Unfortunately, the poor tissue preservation again precludes detailed observations. Th e thin sub-periosteal primary cortex contains a few longitudinally oriented primary osteons. Otherwise all the bone appears to have been remodeled into dense Haversian tissue.

Description. A cross section in compact bone tissue.
Th e section is irregular in shape, about 8 × 5 mm. It seems that an original external free surface of the bone is preserved. Th e external cortex is characterized by many osteons in a periosteal tissue containing many Sharpey's fi bers. A few osteons appear to be primary, longitudinally oriented and lying comformably within the periosteal tissue. However, many osteons are secondary, as evidenced by their inconformity with neighbouring structures. Deeper in the bone, most osteons are secondary, although some primary tissue, with Sharpey's fi bers, locally form an "intersticial system" between them. Most secondary osteons are well defi ned, with a small diameter Haversian canal in the middle. Very few wide open erosion rooms are observed and there is no generalized superposition of several generations of osteons.
Comments. Th e tissue can be described as "dense Haversian bone" although there are probably more primary osteons than suggested by a quick casual observation and secondary osteon overlaping is not overwhelming. Most secondary osteons would be thus of "fi rst generation", suggesting a not so mature tissue. As frequent in fossil dense Haversian tissues (especially from Hadrosaurs "ossifi ed tendons") numerous short radial cracks unite the peripheries of neighbouring secondary osteons, crossing the cementing line limiting each secondary osteon.

Description.
A smaller cross section in compact bone from the same material as above (82) but slightly thinner, allowing sharper observation, but with no preserved peripheral cortex. A few wide open erosion rooms and some "third generation" secondary osteons are observed.
Description. An irregular and ovoid cross section in a tooth 6 × 3 mm. Most of the section is formed of a highly birefringent tissue, surrounding an oval central structure about 2 × 1 mm. Th e external tissue is entirely non-vascular but densely cellular, and entirely permeated by an extremely large number of thin, slightly undulating and very long radial fi bres. Many circumferential lines, closer to each others towards the periphery, are resolved as associated with cracks in the tissue. Although those lines are artefactual, at least in part, they strongly suggest lines of cyclical deposition. Th e inner tissues are sharply delineated from the external one by a clear discontinuity. Deeper to this discontinuity, a fi rst thin layer is amorphous, isotropic and featureless. Th en a second thicker discontinuity circles the most internal tissue. Th e latter is highly birefringent and surrounds a tiny, central free pulp cavity. Th e internal tissue appears mostly globular, especially close to its periphery. Birefringence underlines both radial and circumferential organizations. Successive circumferential regions of relatively more globular or more inotropic mineralizations (sensu Ørvig 1967) seem to alternate. Elongate, very thin radial fi bres can be observed, perhaps as numerous, but far less obvious, than in the surrounding cellular tissue.
Comments. Champsodelphis Gervais 1848 in Gervais 1848-1852 was erected on material from the middle to the upper Miocene of Europe (Simpson 1945). By comparison with available data (e.g., Schmidt & Kiel 1971) we interpret the cellular, non vascular tissue in the periphery as a very thick coat of cementum. Radial fi bres there would be the anchoring fi bres, akin to Sharpey's fi bres. Th e inner tissue can be interpreted as orthodentine, although of a highly globular variety. Th e thin amorphous coating at the periphery remains puzzling. It appears to be natural and well in situ between cementum and dentin. Its optical properties do not suggest enamel, an interpretation which would also be contradicted by its surrounding by the cementum. Description. A thin section in a large tooth, broken in several fragments, the largest being about 10 × 10 mm. Preservation is poor because of breakage during preparation and later shrinkage of the balsam. It nevertheless allows detailed observations locally. A thin coating of cementum is still located over the dentine in one place, and detached in another. Th e cementum is cellular, non vascular, slightly birefringent, and contains thin radial fi bers. Th e region in contact with dentine diff erentiates into thin alternating layers where the relative amount of cementocytes, fi bers and ground substance diff er. In most other regions, cementum has been extensively altered by post mortem boring organisms. Th e dentine forms a very thick (several millimetres) birefringent orthodentine, with numerous, very long and gently undulating canaliculi, with little change from the periphery to the pulp cavity. At least 40 alternatively darker and lighter «growth bands », parallel to each others and to the dentin surface, are visible (but better expressed in the periphery) in ordinary and polarized light. Th ey are supplemented by at least two "lines of arrested growth" locally forming clearer discontinuities. Th ere is no enamel on the section.
Comments. Macrotherium Lartet, 1837 is the same as Chalicotherium Kaup, 1833. Material from the Gers (Sansan) allowed deciphering the puzzle of the Chalicotheria. Macrotherium was known by postcranial material only, while only skull and teeth material were apparently available for Chalicotherium (both from Sansan). Finally, the discovery of subcomplete skeleton in connection at Sansan by Filhol in 1887 demonstrated the common origin of the cranial and post-cranial material (e.g., Viret 1958: 413-420). In connection with the above story, it is of interest that this thin section of tooth is nevertheless labelled "Macrotherium du Gers". As no tooth section bears a date, it is not possible to conclude that attribution of teeth to Macrotherium (rather than to Chalicotherium) was already done prior to Filhol's discovery (1887). Nevertheless, dates on the thin sections from Macrotherium post-cranial bones in this collection (1878)  Description. A small (10 × 5 mm) cross section in compact bone. Th e natural periphery of the cortex is preserved. Primary bone tissue of periosteal origin is locally preserved close to the bone surface. Elsewhere, it has been eroded away, degraded by post mortem organisms, and replaced in vivo by Haversian tissue. As preserved, primary bone appears to be parallel-fi bered, with few vascularisation. No typical "external fundamental system" is obvious. Most of the bone is composed of dense Haversian tissue, with superposition of several generations of secondary osteons, locally. Th e overall very weak birefringence of the bulk of the osteons suggests that most fi bers are perpendicular to the section.
Comments. Structures in the cross and longitudinal sections agree with the hypothesis that they all come from the same region in the same bone shaft, as also suggested by evidence of sawing planes along some sections. Extreme prevalence of the organization of the bone fi bres along a longitudinal pathway may be tentatively linked to the peculiarities of front arm elongation and likely specialized bio mechanics in Chalicotheres.

Chalicotherioidea
Pernatherium. Th is genus, described by Gervais (1876), is classifi ed by Simpson (1945)  Descriptions. Th e fi ve sections seem to come from the same material, if not from the same bone. Sections 93 and 94 are the only ones bearing anatomical information (calcaneum). Section 90 shows almost identical structures with 93 and 94, and will be described with them; 91 and 92 are diff erent.
Calcaneum sections (93, 94 and, very likely, 90) show cross sections of a rather spongy bone, about 13 × 8 mm in surface, at most. Natural surface of the bone is partly preserved as a fl at to gently convex cortex, ornamented by an extremely rough, spiky structure. Th e cortex contains a few irregular primary osteons embedded in a periosteal tissue dominated by extremely abundant Sharpey's fi bers. Th e fi bers are mostly organized perpendicularly or obliquely to the free surface of the bone. Th ey are also organized as superimposed zones, suggesting bursts or cycles of bone deposition, and are well developed in the bone spikes protruding from the general surface. Erosion-reconstruction of the cortex is demonstrated by numerous secondary osteons. Th e superfi cial cortex merges with a more concellous secondary tissue forming the bulk of the bone. However this tissue can hardly be described as a spongiosa because its trabeculae are thick and occupy a larger volume than the cavities in between. Th e tissue is secondary, irregular, formed of endosteal trabeculae forming more or less typical secondary osteons. Interstitial islands of periosteal bone with extensive amounts of Sharpey's fi bers are nevertheless present deep in the bone, far away from the preserved cortex. Th ere is no evidence of articular surface or endochondral components.
Comments. Th e structure fi ts with an endoskeletal short bone, with little contrast between a compact cortex and a spongy medulla. Th e cortex structure strongly suggests a region of dense, extensive connective tissue attachment, probably the tendon of a powerful muscle. Th e obvious biomechanical involvement of the bone as a whole is also suggested by the general robustness of its trabecular system.
Section 91 is a fairly large (14 × 7 mm) longitudinal section in compact bone. Th e tissue is optically very "active", showing high birefringence even under uncrossed nicols. Most of the bone appears to be primary and parallel-fi bered, with a homogeneous mass reaction under crossed nicols. Th e tissue is permeated by a few longitudinal vascular canals. Some canals are crossed transversally and appear as small primary or secondary osteons. No Sharpey's fi bers or lines of arrested growth are obvious. Towards one end of the section, the tissue becomes very progressively invaded by more and more numerous and large secondary osteons, observed in longitudinal section but slightly obliquely. Ricqlès A. de et al. Th is structure suggests a long bone cortex. Th e longitudinal section would be slightly oblique relative to the bone structure, from a more external region (dominated by primary periosteal bone tissue) to a deeper region (dominated by secondary osteons).
Section 92 is a longitudinal section in compact bone, smaller than 91, triangular in shape with each side about 10 mm. As in 91 and even more, the tissue is highly birefringent. Vascular canals are cut longitudinally and transversely. Th e bone tissue is entirely endosteal, with complex evidences of multiple erosion/reconstruction sequences and secondary osteon-like structures cut longitudinally and obliquely.
Th e simplest interpretation is that this section comes from the same material as 91, cut parallel to it but slightly deeper into the bone in the perimedullar region (endosteal margin), hence only containing endosteal secondary material. Description. Th is puzzling section, almost translucent, appears to have received some staining, perhaps by fuschin. Th e tentative description herewith is purely provisional.
A lateral part of the section appears to comprise cancellous bone. Th is merges in the medial part of the section into a dense tissue permeated by small (vascular?) cavities set in extremely orderly longitudinal rows parallel to each others. Th e tissue also contains numerous tiny cell spaces set regularly in pace with the above mentioned longitudinal rows. Finally, the tissue is entirely permeated by numerous small branching canals, slightly larger in diameter than the supposed cell spaces, but smaller than the putative vascular spaces. Again, this system is orderly set in pace with the general structure of the tissue but runs generally perpendicular to the longitudinal rows.
Comments. Epihippus Marsh, 1877 is a "primitive" horse from Middle to Late Eocene of Western North America, known by fragmentary material (e.g., Viret 1958;Mc Fadden 1992). It is not clear whether this thin section actually comes from this taxon or not.
Th e structure is puzzling; it could be tentatively interpreted as a tangential section into the ossifi ed part of a large tendon at the level of its insertion into bone. Th e numerous branching canals perpendicular to the general longitudinal organization of the tissue are best interpreted as evidence of a massive post mortem invasion by boring micro -organisms, the invasion being spatially constrained by the strongly ordered longitudinal organization of the tendon-like tissue.

Description.
A 10 × 15 mm cross section in compact bone. Cell spaces and their canaliculi are well underlined by post mortem deposition of dark mineral salts, but the whole bone is microfragmented and the numerous cracks hinder observations. Only a small region may represent the external cortex. Th e few primary bone left is very poorly vascularized, already replaced by secondary osteons. Many highly irregular, grey-brown coloured secondary osteons strongly contrast with a more translucent yellowish background of dense bone tissue. Th is background is formed by densely packed irregular bony trabeculae and secondary osteons. Th e structure may thus derive from a compacted coarse cancellous tissue of endosteal origin (Enlow 1963) rather than from a "regular" cortex turned into dense Haversian bone. No remains of calcifi ed cartilage or endochondral ossifi cation could be observed.
Comments.Th is very compact, irregular secondary tissue fi ts the general pattern of "pachyosteos clerosis" common in parts of the Sirenian skeleton (e.g., Ricqlès & Buff rénil 2001). Th e strong diff erential natural colouring of some secondary osteons and other structures underlines the in vivo diff erences in bone matrix mineralisation, the latest (youngest) secondary osteons laid down being less mineralized than the older (more mature) ones that form most of the more translucent "background". Proboscidea 97. "Mastodonte de Sansan, cubitus". Pencil number: 271, ink label: 1938-582. Anat. Comp. 1878.

Description.
A thin bony "scale" probably detached or sawed from a shaft and sectioned transversely. Th e section (9 × 2 mm) is technically good. One side of the section, fl at and perhaps sawed off , is entirely Haversian with small secondary osteons and a few, much larger, irregular ones. Deeper in the section, the bone is entirely formed by larger and larger endosteal osteons, ending into large, broken endosteal trabeculae.
Comments. Th e structure strongly suggests a perimedullar region in the endosteal margin.

Description.
A larger section about 10 × 12 mm. Th e bone, sectioned longitudinally, is compact and displays a complex structure, with large, longitudinally sectioned secondary osteons spreading in a complex of dense endosteal trabeculae.
Comments. Th is longitudinal section matches the bone organization in an endosteal margin.

Description.
A longitudinal section of 19 × 12 mm in bone cortex. Th e section contains two fragments with diff erent structures. One fragment, covered by sediment, shows the free external surface of the cortex on its two sides (because of tangential eff ect of the section in a convex structure). Th is superfi cial cortex is largely monorefringent, and contains a few longitudinally and circularly oriented primary vascular canals, and numerous Sharpey's fi bers. Deeper in the cortex, the structure becomes almost entirely Haversian. Th e external cementing lines of the secondary osteons are clear, as well as their central Haversian canal and its anastomoses branching laterally. Th is is well developed in the second fragment.

Description.
A cross section in cortical bone and endosteal margin. Th e section broke during the technical process, which hinders observations. Th e free surface of the bone and its external-most cortex are lacking. Tiny remains of the periosteal bone tissue can nevertheless be observed locally: they contain longitudinally and circularly oriented Ricqlès A. de et al. small primary osteons embedded in an apparent lamellar matrix containing Sharpey's fi bres. No fi bro-lamellar organization of the laminar or plexiform subtypes could be deciphered. Most cortical bone is densely Haversian. Th is secondary cortex gently merges, very progressively, into a well-developed endosteal margin, with larger and larger endosteal osteons and large secondary endosteal trabeculae. Description. A small (4 × 6 mm) longitudinal section that apparently comes from an endosteal margin subject to diff use erosion. However, a small region is formed by periosteal tissue with Sharpey's fi bers, and possibly a "bright line" of the laminar bone pattern. Th e numerous osteocytic lacunae and canaliculi are of exceptional sharpness. Description. A small (7 × 5 mm) cross section in cancellous bone. Th e tissue is formed of complex endosteal trabeculae, with extensive evidence of erosion/reconstruction cycles. In the thickness of some trabeculae, typical smaller secondary osteons were laid down. Between the network of trabeculae, a rather regular pattern of circular marrow spaces is produced.

"Dinotherium
Comments. No remains of either periosteal bone or tissues of endochondral origin are left within the trabeculae, which suggests a perimedullar spongiosa from the bone shaft.

Description.
A large (about 20 × 10 mm) longitudinal section, in poor condition. Both bad post mortem preservation, technical and curation problems (the balsam is badly cracked) prevent meaningful comments. Th e cover slip was rectangular and apparently disapeared. Observation at high magnifi cation confi rms the generalized invasion of the bone tissue by post mortem boring organisms.

Description.
A section of 10 × 5 mm in a homogeneous dentinal tissue. Macroscopically, the whole section is divided by numerous wide undulating lines, roughly parallel to each others. Th ey can be observed only in polarized light (crossed nicols) and are produced by the regular ordering of the dentinal tubules over great distances. Th e numerous tiny dentinal tubules appear to be sectioned transversally and somehow obliquely. Th e tissue is pervaded by numerous black spots with stellate, irregular limits, which appear to be metallic oxides.

ASSESSMENT AND CONCLUDING REMARKS
Paul Gervais' comparative collection of thin sections from fossil bone and teeth retains a current interest in many ways. First, it documents Gervais' own curiosity and attention towards a subject -comparative hard tissues histology -still relatively new and little known at his time, where he obviously hoped to fi nd new tools and possibilities to identify and diagnose fossil material. His similar interest for the comparative microstructures of reptilian and avian eggshells during the same period (Taquet 2001) supports this view. Th e fact that Gervais realized that the very limited knowledge available at his time on eggshells structure precluded reaching fi rm taxonomic conclusions (Taquet 2001) may have extended to his assessment of the situation regarding bone and teeth histology. Th is could explain why Gervais ultimately made little use of his collection of thin sections at the time (Gervais 1875a, b; see also Crémière 2004 for more general comments on microscopical researches at MNHN during the last decades of the 19th century).
As for the eggshell collection, in most cases, Paul Gervais took great care of the orientation of the sections relative to the structures: very often, longitudinal and cross sections were performed from the same specimen. For some 70% of the sections from mammalian limb bones, he focused on the ulna (cubitus), and thus seems to have sought some sort of standardization. It is unknown whether he had special reasons (other than operational) to concentrate on that bone.
In line with Paul Gervais's main scientifi c interests in paleontology, the mammals occupy a prevalent place in the collection (almost 80%). Amphibians (9%), Reptiles (11%), and especially fi shes sensu lato (less than 3%) occupy a very minor position, exemplifying a long tradition in French vertebrate paleontology, where "lower vertebrates" were somehow restricted to the inferior condition that the expression implies (Ricqlès 2006). Th is would be accordant with the general "progressionnist" views of life that the early French evolutionary paleontologists, notably Albert Gaudry, promoted at the time (e.g., Gaudry 1878). At best, non mammalian vertebrates appear here as a small comparative background. A few dinosaurian bone sections, however, document Paul Gervais' relationships with Matheron (and others) and his likely hopes to decipher histologically the signifi cance of the large bone and eggshell fragments from the Cretaceous of Southern France. Within Mammals, the balance among the main Orders is also interesting, even puzzling: while the Carnivores and Primates are lacking, an enormous percentage (45%) is devoted to the South American Mammals, Notongulates and Edentates, sensu lato. Among Ungulates the prevalence of Perissodactyls over Artiodactyls may also raise questions. Of course, all the above assumes that the present survey of extant sections is a faithful refl ection of the original collections, which is still unknown (see Introduction). Indeed, Gervais's short histological publications of 1875 focus on a specialized issue: hyperostosis, which is dealt with in a well balanced way among the various vertebrates groups (including fi shes). Th e sections described by Gervais in that paper do not belong to the list of sections currently found and described herewith.
For all these reasons, it seems that the collection was built up following some circumstancial constraints rather than trying to establish a well balanced and general systematic coverage of mammalian (and vertebrate) bone and teeth histodiversity. Th e collection may mostly refl ect Gervais' own long term focal interests (e.g., 1855, 1873, 1874), but also special opportunities and questions, such as problems arising from the discoveries in Sansan (Gers) site, and from the availability of the fossils (notably those from South America) to become permanent displays in the Galerie de Paléontologie at MNHN (opening in 1898), etc. Similar opportunities may have arisen from the discoveries by Matheron and others of dinosaurian remains in the Cretaceous of Provence-Languedoc (e.g., Taquet 2001). However, Gervais' willingness to acquire histological material for comparison from England is a testimony of the depth of his involvement in such researches.
From a histological point of view, many sections were of excellent technical quality, at least initially. After more than 120 years, most of them are still readily usable. Problems were obviously encountered with heterogeneous materials, such as teeth, Ricqlès A. de et al. which often suff ered dislocation due to the lack of an effi cient embedding medium before sawing and grinding. Nevertheless, even in such cases, histology can be deciphered. A high percentage of the thin sections (about 25% of mammalian bones and teeth) exhibit bad to very bad post mortem preservation, a signifi cant ratio that may have somewhat discouraged Paul Gervais to extend his paleo histological inquiry further. Th e great prevalence of dense Haversian bone tissue in many sections from Mammals, combined with little evidence of the actual histodiversity of primary (periosteal) bone deposits in this group, may have also induced Gervais to relinquish from initial hopes of taxonomic determination from bone histology.
Since Gervais' histological endeavors, a full century has elapsed, during which a signifi cant body of knowledge on the comparative histology of hard tissues in mammals was collected (e.g., Foote 1916, Amprino & Godina 1947, Enlow & Brown 1956-58, Schmidt & Kiel 1971. It is remarkable that his collection of thin sections completes our knowledge of hard tissue histology in some fossil mammalian groups, on questions that still remain poorly documented in modern literature, as far as we know. Th is clearly exemplifi es the patrimonial and scientifi c value of historical comparative paleohistological collections.