Eospondylus cf. primigenius (STÜRTZ)

Hotchkiss, Frederick H. C., Prokop, Rudolf J. & Petr, Václav, 2007, Isolated Ossicles Of The Family Eospondylidae Spencer Wright, 1966, In The Lower Devonian Of Bohemia (Czech Republic) And Correction Of The Systematic Position Of Eospondylid Brittlestars (Echinodermata: Ophiuroidea: Oegophiurida), Acta Musei Nationalis Pragae Series B 63 (1), pp. 3-18 : 8-16

Species Eospondylus cf. primigenius (STÜRTZ)

Text-figs 3-5, 8 , 9A, 9B , 11.

H o r i z o n: Loděnice Limestone, Třebotov Limestone, Lower Devonian

M a t e r i a l: NM L 36903, NM L 36904, NM L 36905, NM L 36910, NM L 38365, NM L 38366, 10 unnumbered specimens from Dalejan and 8 unnumbered specimens from Pragian

O c c u r r e n c e: Praha – Klukovice, Červený lom quarry (Pragian); Holyně, Prastav quarry (Dalejan)

Description of vertebrae

The complex shape and the general proportions of the ossicles are best understood by examining the SEM stereopair dorsal and ventral ( Text-fig. 3), proximal and distal ( Text-fig. 4), and left and right ( Text-fig. 5) views of each vertebra. The vertebrae carry some damage, especially the fragile cupola structures for the tube feet are broken or missing. Of twenty-three vertebrae, two are half-vertebrae and twenty-one are entire. Traces of the suture between the two firmly-joined halves are evident particularly on the proximal and distal surfaces. The vertebrae are pierced lengthwise by the auluroid canal for the radial water vessel of the water vascular system. The following features are useful to orient the vertebrae. The dorsal surface has a median dorsal cleft. The ventral surface is flat and plastron like and has distal lateral cupola structures for the tube feet. The proximal surface has a prominent unpaired dorsal knob and the opening of the auluroid canal is a deep funnel. On the distal surface the auluroid canal is concentric with the zygosphene knob (hence like a spout), and the zygocondyles look like raised wings of a bird (eagle).

36903 L 36904 L 36905 L

DORSAL VENTRAL

Under surface of vertebrae ( Text-fig. 3): The portion of the under surface of the vertebrae that corresponds with the floor of the ambulacral groove is essentially flat and plastron-like. This part of the ossicle is exposed in the normal outward movement of the lateral plates in articulated specimens. A shallow indent with abundant stereom perforations is present on the mid-line of the plastron. Proximally the perforations define a nearly triangular zone, widest at the proximal edge of the undersurface, and narrowest near the mid-length of the vertebra where indented lateral branches of perforate stereom lead toward the location of the lateral arm plates and the cups for the tube feet. The midline triangle of perforate stereom divides the proximal undersurface of the vertebrae into left and right islands of stereom that lack obvious perforations. The proximal edge of these islands extends as a lobe that partly hides the ventral muscle fossae of the proximal surface. The lateral branches of perforated stereom are succeeded distally by a small area of stereom that lacks obvious perforations. This area is contiguous with a spout-like structure that contains the radial water vessel (auluroid condition). These features are visible in the ambulacral groove of some articulated E. primigenius .

At the mid-length of the vertebra the lateral branches of perforated stereom form a transverse belt. Disto-laterally to this belt the width of the vertebrae expands into the cupola structure of the cups for the tube feet. The cupola structures project from the sides of the vertebrae and are intact in specimen NM L 36905. The cupola structures are broken off on one side of NM L 36904 and off both sides of NM L 36903. Likewise they are broken off both sides of the vertebrae illustrated by Boczarowski (2001, figs 9L, 9M). Although broken off, nevertheless a part of the rim of the cupola is evident in NM L 36903 and in Boczarowski’s Text-fig. 9M . In articulated E. primigenius the cupola structures normally are concealed by the lateral plates. Specimen NM S 4766 is an exception in that a few lateral plates are displaced and expose the cupola area ( Text-fig. 6 ).

The lateral ridges (the proximal and distal borders of the wings of the vertebrae) converge ventrally with the rim of the cupola such that the proximal lateral ridge is seen from the under surface. The cupola hides the distal lateral ridge. The proximal lateral ridge projects significantly from the side of the vertebra in NM L 36905, but due to breakage it projects only slightly in NM L 36904, and does not project into view in NM L 36903.

The under surface view of the isolated vertebrae includes a profile view of the proximal and distal articulation and muscle insertion features which, although prominent, are not part of the under surface. The articular processes strongly project beyond the plastron and cupola regions. At the distal border can be seen: the conspicuous spout-like zygosphene (unpaired ventral knob of distal surface) that contains the radial water vessel, deep paired hollows (the ventral muscle fossae and the pits for the zygapophyses of the conjugate surface are quite confluent), and the zygocondyles (paired knobs of distal surface). At the proximal border can be seen the ventral muscle fossae, the zygapophyses (paired knobs of proximal surface) and the epanapophysis (unpaired dorsal knob of proximal surface). At the distal border of the undersurface of the vertebrae exposed in the ambulacral groove of some articulated E. primigenius ( Text-fig. 6 ) can be seen the recesses of the ventral muscle fossae, the spout-like zygosphene, and even evidence that the zygosphene is penetrated by the auluroid canal. Likewise, at the proximal border can be seen part of the funnel-like ventral socket (the reentrant between the zygapophyses) and parts of the surfaces for the attachment of the ventral intervertebral muscles.

Upper surface of vertebrae ( Text-fig. 3): The halves of vertebrae are solidly fused, without visible suture. The stereom in the aboral groove and on the wings of the vertebrae has abundant large perforations. The aboral groove (median dorsal cleft) is deeply sculpted and broadest at its median part. In NM L 36905 the aboral groove has a lenticular shape and looks like a median pore, but it is only a deep hollow. The upper and lateral part of the vertebra between the insertions for the dorsal longitudinal muscles is the wing of the vertebra. In all specimens the wing of the vertebra is widest at the lateral ridges. Between the lateral ridges the wings are deeply excavated and vertical-sided to the level of the circular cupola. In NM L 36905 the excavation goes more deeply into the ventrolateral area, making a ventrolateral pit just above the cupola. The upper surface of the cupola has a conspicuous hollow. The proximal and distal lateral ridges are confluent with the margin of the cupola. In some articulated E. primigenius (ROM 11474, WU 860033) the prominent spine-bearing ridge of the lateral plate lies precisely over the distal lateral ridge, and we believe that this is the undisturbed position. In BMNH E3357 the lateral plate appears to have slipped off the lateral ridge and into the concavity of the side of the vertebra ( Spencer 1925, text-Text-fig. 206).

The proximal and distal articular processes strongly project beyond the lateral ridges of the vertebrae. This was observed also by Boczarowski (2001, p. 38) and written into his diagnosis of Eospondylus ingens . The proximal processes project considerably more, and the distal processes project less. In NM L 36905 upper surface view the distal processes project very little beyond the lateral ridge, and this is true also in a vertebra of E. ingens interpreted by Boczarowski (2001, Pl. 1, Text-fig. 17) as coming from the middle part of the arm. At the proximal border in all specimens can be seen the conspicuous epanapophysis (unpaired dorsal knob of proximal surface) and the zygapophyses (paired knobs of proximal surface). The articular processes at the distal border are the epanotreme (unpaired dorsal receptacle that lodges the epanapophysis) and the zygocondyles (paired knobs of distal surface, but not looking like knobs in this upper surface view). In articulated E. primigenius the upper surface of the vertebrae is not covered by any plating. The peg on the proximal surface (the epanapophysis) is lodged in a shoulder-socket on the distal surface (the epanotreme) when the vertebrae are in close contact (but in many specimens successive vertebrae are not in close contact). The space for the dorsal intervertebral muscles is fairly large as a consequence of the setback of the lateral ridges from the extremities of the articular processes. The setback for the muscles is greatest in NM L 36905 and involves a small curl in the lateral ridge where the ridges converge behind the epanapophysis.

Proximal surface of vertebrae ( Text-fig. 4): The proximal articulating surface presents a strong image of a central triangular area in the foreground and a peripheral region that is the wing of the vertebra in the background. The shape in NM L 36905 is like a tall isosceles triangle with ventral base, dorsal vertex and rounded angles. In NM L 36903 the triangle is almost equilateral. The triangular area contains all the articulating knobs and pits, the canal for the radial water vessel, and the ventral intervertebral muscle fields. The base of the ‘triangle’ is a profile of the under surface of the vertebra. The profile in NM L 36905 shows a shallow dihedral angle. The profile in NM L 36903 and NM L 36904 shows a shallow ventral groove. The articulation knobs of the proximal surface are the epanapophysis (unpaired dorsal knob of proximal surface) and the zygapophyses (paired knobs of proximal surface). The articulating pits (sockets) of the proximal surface are the zygotreme (unpaired ventral socket of the proximal surface) and two unnamed dorsolateral sockets. The dorsal knob is shaped somewhat like a square that is standing on its diagonal. In this orientation it contributes two articular facets, each at forty-five degrees to the horizontal (following the insight of Bray 1985). The vertical suture of the two fused ambulacral plates is plainly visible on the dorsal knob. Looking a bit like an inverted flask, the dorsal knob is connected below to a median saddle ridge which extends to the opening of the auluroid canal. The canal opening is at the bottom of a deep depression that serves as a socket (zygotreme) for the ventral knob of the conjugate surface. Paired ventral knobs (zygapophyses) are on either side of the canal opening. These angular knobs are somewhat pyramidal with three very steep surfaces: one surface faces toward the dorsolateral paired sockets, one faces toward the ventral intervertebral muscle fields, and one faces toward the radial water vessel. In the perimeter space between the dorsal knob and the paired ventral knobs are the dorsolateral sockets for the horizontal hinge knobs (zygocondyles) of the conjugate surface. Depressions for the ventral muscles are situated at the lower outer corners of the triangular area; the fossae are deeply excavated and of moderate cross section. The dorsal intervertebral muscle fields lie outside of the triangular articulating surface and at a more distant plane (the wing of the vertebra).

36903 L 36904 L 36905 L

PROXIMAL DISTAL

P o i n t o f d i s c u s s i o n: In Eospondylus , at the proximal end of the vertebrae the insertion surfaces for the dorsal intervertebral muscles are at a more distal plane than the ventral muscle insertion surfaces. This is the reverse of the relative locations of these surfaces in the Order Ophiurida (e.g. see Palaeocoma in Hess 1962, p. 617, Text-fig. 7 , lateral view). This difference applies only to the proximal end of each vertebra. The relative location of muscle insertion surfaces at the distal end of the vertebrae is the same in Eospondylus and in the Order Ophiurida . At the distal end of the vertebrae the insertion surfaces for the dorsal intervertebral muscles are at a more proximal plane than the ventral muscle insertion surfaces.

Distal surface of vertebrae ( Text-fig. 4): The articulation processes of the distal surface present the general impression of a bird image (“vogelähnliches Gebilde” of Hess 1962, p. 615). The zygocondyles (paired dorsal knobs of distal surface) look like raised wings. The zygosphene (unpaired ventral knob of distal surface) looks like a bird body suspended beneath the wings. The upper surfaces of the two zygocondyles form a V and acted as paired aboral articular facets (analysis of articular facets follows the insight of Bray 1985). The lower surfaces of the two zygocondyles are configured as / and \ and acted as paired oral articular facets. The outer surfaces of the zygocondyles are steep flat surfaces that bordered the dorsal longitudinal muscles. The large auluroid canal opening is subcircular, is located in the lower half of the vertebra, and is concentric with the zygosphene knob. The canal opening is surrounded by a raised rim. The articulating pits of the distal surface are the epanotreme (unpaired dorsal socket) and unnamed paired ventrolateral pits that merge imperceptibly with deep fossae for the ventral longitudinal muscles. The muscle fields for the dorsal longitudinal muscles are the wings of the vertebra.

Lateral surfaces of vertebrae ( Text-fig. 5): Both sides of NM L 36903 and NM L 36904 have similar appearances, but only the left side of NM L 36905 is used for description because the right side is damaged and has a substantially different appearance (major loss of proximal lateral ridge). The side view dorsal profile of NM L 36905 is a swayback shape, and consequently this vertebra has a lower profile in its midregion than do the vertebrae NM L 36903 and NM L

36903 L 36904 L 36905 L

LEFT SIDE RIGHT SIDE

36904. This swayback low profile is even more strikingly presented in NM L 38365 (elongate swayback version of standard type). The lateral surface is divided into three successive regions. At the ends of the vertebrae are the exposed proximal and distal articulating surfaces. The central region is the wing of the vertebra, ending at the proximal and distal lateral ridges. The wing extends from the dorsal surface to the ventral edge, ending where the cupola is located. The wing of NM L 36905 is broader than in NM L 36903 and NM L 36904, whereas the dorsal nose of the proximal articulating surface is more prominent in NM L 36903 and NM L 36904 than in NM L 36905. The appropriateness of the term dorsal nose for the epanapophysis is evident in side view. The proximal articulating surface, described from upper part to lower part, is composed of a shoulder leading to the dorsal nose, a reentrant below the nose and also at this level a deep dorsolateral pit, a projection that is the zygapophysis, a deep reentrant that leads to the fossae for the ventral longitudinal muscles, and a final ventral projection that contains the muscle fossae. The zygapophyses project as much (NM L 36905) or slightly more (NM L 36903, NM L 36904) than the dorsal nose projects. The profile of the distal articulating surface has three parts. Beginning at the upper surface, the profile slopes linearly away from the distal lateral ridge; the next part is a vertical profile at midlevel; the bottom part slopes back to the distal lateral ridge at the level of the cupola. The presented surface is uniform and flat. The flatness and the simple profile contrast with the complexity of the epanotreme, the zygocondyles, and the ventrolateral pits and fossae that are behind the top, middle and lower portions of this surface. The zygosphene (unpaired ventral knob pierced by the auluroid canal) is hidden from view, or inconspicuous or visible depending on the tilt of the specimen. The wing of the vertebra is deeply excavated just above the cupola. The shape of the wing of the vertebra is outlined by the lateral ridges. The tilt of the proximal lateral ridge toward the posterior cupola is particularly distinctive.

Variation of the vertebrae: Differences between the swayback lower profile vertebra NM L 36905 and the more block-like NM L 36903 and NM L 36904 are recorded throughout the description. Such differences may indicate that the vertebrae came from different parts of the arm ( Boczarowski 2001), or that they may represent variation among specimens, or some of both. The elongate swayback vertebra NM L 38365 is probably from the distal region of an arm where segments are elongate. The block-like NM L 36903 and NM L 36904 are probably from the proximal or middle part of an arm. The swayback vertebra NM L 36905 is more equiproportioned than elongate and this suggests that it is a transitional morphology, perhaps coming from a part of the arm that is between the midpart and the distal part of the arm. Specimen NM L 38366 is block-like but of very small size, probably indicating the proximal or middle part of the arm in a small growth stage individual. On the other hand some of the variability may represent phenotypic variation in the population ( Text-fig. 1 ), the reasons for which can vary. For example the living ophiuroid Ophiocoma echinata has a more robust form with relatively shorter arms in areas exposed to high energy waves in the swash zone ( Bray 1981). Lehmann (1957) recognized E. primigenius var. compactus as having arm structure different from typical E. primigenius . Measurements (length x width) obtained from Národní Museum SEM images print- ed with a 500 micron scale (the 100 micron scale in the Indiana University images is not as well suited): NM L 36903 1.46 mm x 1.23 mm; NM L 36904 1.48 mm x 1.17 mm; NM L 36905 1.25 mm x 1.25 mm.

Description of lateral arm plate

The one lateral plate NM L 36910 ( Text-fig. 8 , Text-fig. 9 ) is assigned to Eospondylus based on shape comparisons with lateral plates in articulated E. primigenius specimens ( Text-fig. 7B ). Especially similar are the prominent vertical ridge, the outward flair of the surface that leads to the vertical ridge, the blunt pointed denticles that separate the spine sockets, the count of seven vertical spine sockets, the lack of a distal tongue (successive laterals of E. primigenius do not overlap), and the small size of the slit-like groove spine sockets which are five in number. The height profile leaves uncovered part of the side of the arm vertebrae, while ventrally the plate wraps partly under the arm. This lateral plate differs from the proximal, middle and distal lateral plates of E. ingens which have a distal tongue ( Boczarowski 2001, figs. 9 N-Q). This lateral plate is more similar to the lateral plates of Furcaster aequoreus (especially Boczarowski 2001, Text-fig. 11 B ) but the Furcaster lateral plates have a narrower and more pointed proximal end.

Analysis and conclusions

These vertebrae compare very well with the exposed parts of the vertebrae of Hunsrück Slate E. primigenius . In particular they have the identifying characteristic of a medi- an dorsal cleft that deeply indents the upper surface of each vertebra and deep hollows on the dorsolateral surface. With the exception of the elongate distal vertebra NM L 38365, the amount of exposed proximal articulating surface in lateral view is much smaller in the Barrandian ossicles than in the holotype of E. ingens from the Holy Cross Mountains ( Boczarowski 2001, Text-fig. 9L 3). We identify the Barrandian vertebrae as Eospondylus sp. , cf. E. primigenius .

The conclusion that articulated E. primigenius did not have under arm plates ( Spencer 1925, p. 311) is supported by the present study. The detailed correspondence of the underside of the isolated vertebrae with the ossicles exposed in the ambulacral groove of articulated E. primigenius identifies the ossicles in the ambulacral groove as vertebrae that are not obscured by under arm plates. Even the cupola area of the vertebrae is exposed in rare articulated specimens in which a few lateral plates are sufficiently displaced ( Text-fig. 6 ). In addition, the observation that the side arm plates were movable and could either cover or uncover the ambulacral groove, and the presence of groove spines on the ventral edge of the side arm plates ( Text-fig. 10 ), are consistent with a lack of under arm plates.

The architecture of the articulating surfaces of Eospondylus is both zygospondylous and auluroid ( Text-fig. 11 ). Weller (1930) described isolated vertebrae with this articulation type but he was not able to identify them to family or genus. Identification came when Haude (1982, Text-fig. 3) and Haude et Thomas (1983, Text-fig. 7 ) found vertebrae of this type plan in partially disarticulated specimens of Furcaster n. spp. (Family Furcasteridae ). Haude (1982, p. 14) and also Boczarowski (2001, p. 29) recognized several vertebrae illustrated by Becker et Weigelt (1975) as belonging to Furcaster . Most recently Boczarowski (2001) described and named Furcaster aequoreus , F. cataphractus and Eospondylus ingens based on isolated vertebrae of this articulation type. In this architecture, (1) the articulations are a topological match to the trispondylous peg-and-socket articulations seen in many Ophiurida , (2) the channel for the radial water vessel is completely enclosed within the vertebra, and (3) the placement of the canal for the radial water vessel is in the center of the zygosphene knob and zygotreme pit. Thus the zygosphene with its hole for the radial water vessel is like a spout or nozzel, and the zygotreme is like a funnel. So far, this type of arm vertebrae with zygospondylous articulations and centered auluroid canal is known only in the Furcasteridae and the Eospondylidae .

P o i n t s o f d i s c u s s i o n: A vertebra figured by Becker et Weigelt (1975, plate 4 figs. 6a-e) has some eospondylid features but differs from typical Eospondylus and Furcaster vertebrae. The orientation of the views should have been stated as follows: 6a right side, 6b dorsal, 6c ventral, 6d distal (posterior), 6e proximal (anterior). This vertebra may possibly belong to the family Eospondylidae .

The proximal vertebrae of Furcaster cataphractus have many features in common with Eospondylus vertebrae and also significant differences. In proximal and distal end views the articulation processes share significant resemblances. In oral view the expanded cupola for the tube feet are quite similar (compare Boczarowski 2001, Text-fig. 13). However in F. cataphractus the aboral groove is approximately straight-sided and situated between narrow sharp ridge borders (while it is lens-shaped and situated between wide rounded borders in Eospondylus ). In F. cataphractus the distal wings are so close to the end of the vertebrae that in side view the articulation processes are mostly hidden (while in Eospondylus alar ridges set back from the articular surfaces which are visible in side view). In F. cataphractus the sides of the vertebrae are broadly concave (but in Eospondylus there is the nonhomologous deep excavation in the wing area just over the cupola). So far the Barrandian isolated vertebrae of Furcaster and of Eospondylus have been easy to separate.

The distal surface of a proximal vertebra of Palaeocoma gaveyi figured by Hess (1962, Text-fig. 7 ) also has the three features of being (1) zygospondylous, (2) auluroid, and (3) with the hole for the radial water vessel at the center of the zygosphene. The Hess vertebra, which is not typical of the rest of the arm vertebrae, suggests that the manner of transition from a centered auluroid canal to an open ventral channel for the radial water vessel might usefully be studied in the arm of P. gaveyi . Spencer (1925, p. 251) argued that there is great plasticity in the relation of the radial water vessel to the substance of the vertebra, and therefore he argued that the auluroid condition is not a basis for erecting a Class Auluroidea. His view is confirmed by Trichaster (an euryalid with hourglass articulations) in which the radial water vessel is enclosed in a ventral canal that is not part of any articulating surface ( Mortensen 1933).

Although sharing generalized topological structure with the vertebrae under discussion, the vertebrae of Hallaster and Lapworthura as illustrated by Spencer (1925, figs 178, 205) are too different to be considered at length here. In Lapworthura each half vertebra has its own functional dorsal nose [not so in Hallaster in which the vertebral halves share in forming a single dorsal nose]. In both Hallaster and Lapworthura the ventral nose is scarcely evident and also the hole for the auluroid canal is very small, inconspicuous, and near the ventral surface.

As is well known, the arms of the Euryalida typically coil ventrally and may branch. In particular, the vertebrae of the Euryalida typically articulate by broad hourglass-shaped surfaces ( Mortensen 1933; Spencer et Wright 1966, p. U90). Although Eospondylus was capable of ventral arm coiling, aboral arm flexure, and extreme horizontal arm bending, we have shown here that the vertebrae of Eospondylus do not have hourglass articulations and are not intermediate between hour-glass type and zygospondyline type. In addition, articulated body fossils of Eospondylus do not have radial shields, do not have genital plates, do not have oral shields, do not have adoral shields, and do not have buccal under arm plates, whereas all of these are present in the Euryalida . We argue that the total evidence against classifying Eospondylus in the Euryalida is convincing.

From its vertebrae Eospondylus is almost a Furcaster . We classify the Eospondylidae next to the Furcasteridae in the Suborder Zeugophiurina , in the Order Oegophiurida . This result is much closer to the classification of Spencer (1925) than to the classification of Spencer et Wright (1966). Spencer (1925) placed Eospondylus in the family Hallasteridae because the long spines are “set at a distinct angle to the direction of the arm-length” (p. 309), and he placed Furcaster in its own family Furcasteridae because the arm spines are parallel to the arms. On the other hand, he considered that the vertebrae and side shields of Hallaster , Lapworthura and Eospondylus are so similar to those in Furcaster that he placed the family Hallasteridae next to the family Furcasteridae . Certainly more work has to be done to elucidate the morphology of all these genera and to use cladistic methods to formulate hypotheses of relationships among them. Work of this sort is being done by Glass (personal communication, June 2005). In an abstract for the 2005 Geological Society of America meetings he reports that separation of the Furcasteridae and the Eospondylidae is not supported.

A conclusion from this analysis is that Eospondylus is not the first occurrence of the Order Euryalida . It follows that neither is the sister taxon Kentrospondylus . According to the classification of Spencer et Wright (1966) the candidate after Eospondylus and Kentrospondylus for the first occurrence of the Order Euryalida defaults to Onychaster . However, retaining Onychaster in the Euryalida is not likely to withstand critical reevaluation because it is known that Onychaster has an auluroid canal and does not have hourglass articulations ( Spencer 1927, text-Text-fig. 217). The classificatory importance that Spencer (1927, p. 333, 325) attached to a single Onychaster specimen with a branched arm is greatly diminished by the lack of any similar occurrence despite many examined specimens. The USNM Onychaster specimen from Indian Creek with the branched arm has been lost track of ( Owen 1965, p. 557). A single instance of a branched arm could be an accident of regeneration. The first undoubted euryalids are post-Triassic.

Figures

Text-fig. 8. Eospondylus cf. primigenius (STÜRTZ) "Červený lom" quarry near Praha-Klukovice, Loděnice Limestone, Lower Devonian, Pragian, NM L 36910, x 25. Left lateral plate, outer view. The surface leading to the vertical ridge flairs outward. The height profile leaves uncovered part of the side of the arm vertebra.

Text-fig. 9A. Eospondylus cf. primigenius (STÜRTZ) "Červený lom" quarry near Praha-Klukovice, Loděnice Limestone, Lower Devonian, Pragian, NM L 36910, x 40. Left lateral plate. Inner surface view from below. The spine ridge faces distally (posteriorly) and has blunt pointed denticles. Anterior and parallel to denticled part of spine ridge is internal rounded ridge. Ventrolaterally ossicle has slit-like sockets for groove spines.

Text-fig. 11. Eospondylus cf. primigenius (STÜRTZ) “Prastav” quarry at Praha-Holyně, Třebotov Limestone, Lower Devonian, Dalejan, NM L 36905, x 65. Overlay of proximal and distal articulations. Upper photo is proximal surface with distal bird-like articulation knobs superposed in ink. Lower photo is distal surface with proximal articulation knobs superposed in ink. The architecture of articulation surfaces is both zygospondylous and auluroid. This architecture occurs also in vertebrae of Furcaster and indicates that families Eospondylidae and Furcasteridae are closely related.

Text-fig. 6. Eospondylus primigenius (STÜRTZ) Bundenbach, Eschenbach-Bocksberg quarry, Lower Devonian, Lower Emsian (Zlichovian), Hunsrück Slate, NM S 4766, x 12.5. Undersurface of arm segments in Hunsrück Slate articulated specimen. A few lateral plates are displaced and expose details of vertebrae. The canal for radial water vessel is in center of zygosphene knob and zygotreme pit, like spout and funnel. At distal outer edges of vertebrae is cupola for tube feet. There are no under arm plates.

Text-fig. 7A. Eospondylus primigenius (STÜRTZ) Bundenbach, Eschenbach-Bocksberg quarry, Lower Devonian, Lower Emsian (Zlichovian), Hunsrück Slate, BMNH E3353, x 3. Extreme horizontal arm bending: Arm loop with half-twist in Hunsrück Slate. In the proximal part of the arm the view is of the upper surface, but after the loop the view is of the lower surface. Seven arm segments (six articulations) bend 180 degrees, indicating 30 degrees of bend per articulation.

Text-fig. 1. Eospondylus primigenius (STÜRTZ) Bundenbach, Eschenbach-Bocksberg quarry, Lower Devonian, Lower Emsian (Zlichovian), Hunsrück Slate, specimens PWL 1996 57LS (NMM) top image (A) x 1.5, and PWL 1996 66LS (NMM) bottom image (B) x 1.7. Specimen with rays bent into strong curves. Such flexibility and attitudes of rays differentiate the behavioral ecology of Eospondylus from all other Hunsrück Slate ophiuroids. Delicateness of the top specimen and robustness of the bottom specimen is perhaps the combined result of real phenotypic differences, differences of preserved detail, and differences of distortion of the rock. [courtesy of Herbert Lutz and Thomas Engel; photos by Alexander Glass].

Text-fig. 10. Eospondylus primigenius (STÜRTZ) Bundenbach, Eschenbach-Bocksberg quarry, Lower Devonian, Lower Emsian (Zlichovian), Hunsrück Slate, NM S 4766, x 26. Lateral plates on undersurface of arm segment in Hunsrück Slate articulated specimen. Short thin groove spines on ventral edge of lateral arm plates protect undersurface of vertebra and partially cover ambulacral groove.