Ossirarus kierani Clack & Smithson, 2016

Holotype.

UMZC 2016.3. A single block containing scattered skull and postcranial remains.

Locality.

Ross cliffs, Burnmouth, Scottish Borders Region, Scotland. National grid reference NT 964606.

Horizon.

340.5 m above the base of the Ballagan Formation. CM palynozone, mid-Tournaisian, Mississippian.

Emended diagnosis.

Autapomorphies: tabular elongate triangle forming a conspicuous tabular horn with a convex lateral margin.

Derived characters present in several stem amniotes: tabular-parietal contact; exoccipital separate from basioccipital; multipartite gastrocentrous vertebrae with widely notochordal centra.

Plesiomorphies and characters of uncertain polarity: jugal with extensive postorbital component, with anteriorly placed shallow contribution to orbit; preopercular and intertemporal present; cleithrum with long, narrow, curved stem and expanded dorsal blade; diamond-shaped interclavicle lacking parasternal process; humerus with elongate and oblique pectoralis process comparable with the ventral humeral ridge of elpistostegalians and Acanthostega; brachial foramen piercing posterior edge of humerus at the base of entepicondyle as in Mesanerpeton; radius c. 60 % the length of humerus; neural arches as unfused bilateral halves.

Description.

Skull. General skull preservation. The bones are generally well preserved. They are disarticulated and have drifted apart slightly, so that sutural overlap areas are often very clear. The pre-orbital region is missing, and the lower jaws and other tooth bearing bones are not preserved apart from a fragment of maxilla or premaxilla (Fig. 2). We estimate that the skull was 54 mm long and the preserved region made up approximately two thirds of its length. Apart from the tabular and squamosal, which show some ornamentation, the skull roofing bones are essentially smooth. The lateral line system is represented by pores in the jugal, postorbital, postfrontal and preopercular; there are no open lateral line canal grooves.

Cheek region. Much of the right cheek is preserved. It comprises the jugal, postorbital, an incomplete right squamosal and a preopercular. The left cheek is represented by the posterior part of the jugal.

The jugal is a large bone. It is c. 25 mm long, including an extensive area overlapped by the quadratojugal, and has a maximum depth of c. 7 mm behind the orbit margin (Fig. 2). The orbit margin is shallow and below it the bone is relatively deep and posteriorly elongated. It most closely resembles the jugal of Acanthostega (Clack 2002) and colosteids (Smithson 1982; Hook 1983). It lacks the tall, vertical orbital margin seen in Diploradus (Clack et al. 2016), Pederpes (Clack and Finney 2005) and Whatcheeria (Lombard and Bolt 1995; Rawson et al. 2021). The suture with the lacrimal is vertical and judging by the shape of the orbit margin sits under the centre of the orbit. The area behind the orbit is gently concave. It is unclear if this depression is natural or the result of crushing. The dorsal edge of the bone behind the orbit margin is damaged but it appears to have overlapped the ventral edge of the postorbital. The posterior portion of the jugal bears numerous fine ridges and furrows and was probably overlapped by the quadratojugal. This area is large and represents approximately one sixth of the area of the jugal. A much smaller area of ridges and furrows on the posterodorsal edge of the jugal probably formed part of the area overlapped by the squamosal. The incomplete left jugal is represented by the posterior portion bearing the ridges and furrows of the quadratojugal overlap area.

The postorbital is almost rectangular in outline with a gently concave orbital margin anteriorly. Its sutural contacts with surrounding bones are well preserved. Ventrolaterally, there is a shallow step from the smooth external surface to an area of fine ridges and grooves marking the area of overlap by the jugal. The ridges and grooves continue onto the posterior edge marking the area of overlap by the squamosal. These ridges and grooves are also found at the anterolateral corner of the postorbital where it was overlapped by the postfrontal. The medial margin of the postorbital is damaged but appears to have formed a thin lamina that overlapped the lateral edge of the intertemporal.

The squamosal is incomplete and appears to have broken into several pieces, most of which have been lost. Two fragments make up part of the posterior edge of the bone and a third formed the anterodorsal portion of the squamosal between the jugal and skull roof, behind the postorbital (Fig. 2). The external surface of the bone bears a fine reticulate ornament in contrast with the smooth surface of the jugal.

The preopercular lies behind the jugal (Fig. 2). We initially thought it was part of the quadratojugal but the presence of a lateral line pore and the extent and orientation of the area of sutural overlap, as well as comparisons with the preopecular of Acanthostega (Porro et al. 2015) and Whatcheeria (Rawson et al. 2021), convinced us it is the preopercular. The bone is roughly triangular-shaped. It is c. 7 mm long and c. 6 mm high. Approximately two thirds of the surface is covered with the fine ridges and grooves that mark the area of overlap with the quadratojugal (Fig. 2) with only about one third exposed on the surface of the skull. The exposed area is roughly quadrangular, and bears a single lateral line canal sulcus in the posteroventral corner. One side of the quadrangle forms part of the posterodorsal edge of the suspensorium. The anterodorsal edge formed a suture with the squamosal and the anteroventral and posteroventral areas were overlapped by the quadratojugal. The preopercular is hypothesized to have occupied a position on the edge of the suspensorium, above the quadratojugal, in a similar position to the preopercular in Ichthyostega (Clack and Milner 2015, fig. 8), Pederpes (Clack and Finney 2006) and Whatcheeria (Rowson et al. 2021), rather than forming the posteroventral corner of the suspensorium as found in Acanthostega (Porro et al. 2015).

Skull table. Much of the right side of the skull table is preserved (Fig. 3) and comprises the parietal, postparietal, postfrontal, intertemporal, supratemporal and tabular. On the left, parts of the parietal, postparietal, supratemporal and tabular are preserved.

The parietals have separated along the midline and the left has drifted back relative to the right. The bones are thin and incomplete. The thickened area around the pineal is preserved on both sides. Using information from each bone gives a minimum anteroposterior length of 15 mm. The incomplete lateral edge of the parietal is thin and appears to have had a broad overlapping suture with the bones of the temporal series. On the right, the parietal appears to be partially overlying the supratemporal.

The incomplete postparietals have separated and drifted backwards. Each is poorly preserved with a fractured dorsal surface and little if any true edge around the bones. The right is the more complete and appears to be approximately square in outline. They are much smaller than the parietals and have an anteroposterior length of c 8 mm.

The tabular is well preserved on the right. It is a relatively large, approximately triangular-shaped bone and its surface is ornamented with pits and grooves. The anterior edge, where the tabular meets the supratemporal is straight, the medial edge which contacts the midline bones is convex, the posterolateral edge is slightly concave and extends well beyond the posterolateral corner to produce a prominent tabular horn. There is no evidence of sutural contact with the squamosal, the lateral edge of the bone is gently rounded and smooth. The areas of sutural contact with the midline bones are very clear. Along most of the medial edge there is a shallow step down from the external surface to a broad area of ridges and grooves that would have been overlapped by the midline bones. At the posteromedial corner the tabular is thickened and the ridges and grooves form a sloping shelf which extends around on to the posterior edge. This posterior shelf probably marks the area of contact with the postparietal, while the broad flat area probably formed the suture with the overlying parietal. This arrangement suggests that Ossirarus had a tabular-parietal suture and is the earliest record of this feature in early tetrapods. The incomplete left tabular shows part of the tabular horn and the two discrete areas of sutural overlap along the medial edge. The pattern of ornament is similar to that on the right.

The supratemporal is well preserved on the right. It has separated slightly from the tabular and is partially overlapped by the parietal. It is incomplete anteriorly where it meets the intertemporal. The external surface is smooth. The posterior part of the lateral edge is gently rounded and shows no evidence of sutural contact with the squamosal. The anterior part is incomplete. As on the tabular, the exposed part of the medial edge bears a shallow step down from the external surface to a broad area of ridges and grooves that would have been overlapped by the parietal. On the left, the supratemporal has separated from the tabular. It is incomplete anteriorly and damaged along the lateral edge, but the straight, butt suture with the tabular is preserved posteriorly, together with an area of ridges and grooves on the medial edge originally overlapped by the right parietal.

The intertemporal appears to be present on the right between the supratemporal and postorbital and partially overlain posteriorly by the parietal. It is a relatively long bone, c. 9 mm is exposed, but its width cannot be determined, because of the overlying parietal. It is incomplete with damaged edges. There appears to be a small area of sutural overlap ridges and grooves at the anterior tip of the bone.

The posterior part of the postfrontal is preserved. The slightly concave lateral edge forms part of the orbit margin. The surface of the bone is smooth and shows a number of pores of the lateral line canal system. The thickened medial edge bears the characteristic ridges and grooves of sutural contact with the midline bones and there is a small area of ridges and grooves on the posteromedial edge suggesting it was overlapped by the intertemporal.

Palate. Very little of the palate is preserved. Part of the quadrate ramus of the right pterygoid and the right quadrate are present (Fig. 4).

The quadrate ramus is represented by a number of pieces which have been displaced posteriorly beyond the tabular horn and squamosal, and medial to the jugal and quadratojugal (Fig. 4). The pieces include an anterior portion bearing the conical recess, a central portion with a finished lateral edge marking part of the rim of the adductor fossa, and a posterior portion folded along a crack. This last piece has a ventrolateral part that would have sutured with the quadrate and a dorsomedial part that would have contributed to the medial wall and roof of the adductor chamber, and sutured with the squamosal. All the pieces of the pterygoid, apart from that forming the rim of the adductor fossa, have broken edges. The pieces bearing the conical recess and rim of the adductor fossa have broken along a simple crack and can readily be restored into their relative positions (Fig. 4). The other pieces are more difficult to align.

The surface of the bone behind the conical recess is lightly pitted. The surface of other broken pieces of the pterygoid is smooth, apart from the posterior-most portion, which is striated and probably represents an overlap area with the quadrate. None of the pieces of pterygoid bear denticles.

The right quadrate is preserved in internal view. It is roughly triangular-shaped with a central concavity. The lateral edge appears to be broken rather than sutural, and halfway up the medial edge is a notch, which presumably formed the quadrate contribution to the paraquadrate foramen that pierces the quadrate-pterygoid suture in some early tetrapods (Beaumont 1977, p. 52: Clack 2003, p. 488). The ventral edge is unfinished and forms the articulating surface with the articular of the lower jaw. The articulating surface is c. 8 mm long, well ossified and has a complicated shape. It is superficially screw-shaped with the axis running along the length of the articulating surface from the lateral to medial edges. It starts on the lateral edge as a ridge, followed by a furrow, and then a larger rounded ridge, followed by a deeper furrow and terminating on the medial edge with a rounded ridge.

Despite its relatively small size (skull length c. 54 mm) the degree of ossification of the quadrate and the form of the sutures suggest that this was a mature individual.

Braincase. The only part of the braincase that is preserved is a small dumb-bell shaped bone lying behind the right tabular which we interpret as an exoccipital (Fig. 2). For a while we debated whether it might be a stapes, but eventually concluded that it is more likely to be part of the occipital arch.

The bone is c. 6.5 mm high with expanded ends. The end nearest to the tabular is considered to be the dorsal end, the exposed surface is the posterior side of the bone and it is interpreted as the left exoccipital. The dorsal surface is slightly damaged while the central portion is covered with smooth periosteal bone and is pierced on the lateral side by a foramen for the hypoglossal nerve (XII). The ventral end is a triangular-shaped area of unfinished bone which probably formed part of the occipital condyle for articulation with the atlas vertebra. Above and alongside the unfinished area the medial edge is gently curved and formed part of the boundary of the foramen magnum.

Restoration of the skull. The preservation of the skull of Ossirarus is unusual. The separation of the individual bones and exposure of the sutural overlap areas is rare and the result of its unusual preservation (see below: Discussion). This displacement of the bones has added an extra challenge to the preparation of a reconstruction of the skull (Fig. 5). Here, the patterns of sutural overlap revealed in μct scanning studies of the skulls of Acanthostega (Porro et al. 2015) and Whatcheeria (Rawson et al. 2021) have been used as a guide to the relationship between individual bones. We have also tried to take account of the incomplete preservation on some bones like the parietal and jugal, where the areas of overlap are thin and have been damaged and where the full extent of the bone is not preserved. Fig. 5 A shows the relationships of the preserved bones on the right side of the skull (mirrored on the left) in the horizontal plane, with the sutures shown as thick lines and the areas of overlap shown as thinner lines. Fig. 5 B is a partial reconstruction of the skull in dorsal view, based on a model prepared by folding a tracing of the bones in the horizontal plane over a moulded block of plasticine.

In Fig. 6 we compare the reconstruction of the skull of Ossirarus with those of a representative sample of tetrapods from the Upper Devonian and early Carboniferous. All are drawn to the same scale. In comparison with many early tetrapods, Ossirarus was relatively small. Upper Devonian tetrapods were typically quite large animals with some exceeding one metre in length (Clack and Milner 2015). Many early Carboniferous tetrapods were equally large. The whatcheeriid Pederpes from the Ballagan Formation near Dumbarton in Scotland was approximately one metre long (Clack and Finney 2005), and Crassigyrinus, which may be represented at Burnmouth by an incomplete lower jaw (Clack et al. 2018), attained a length of nearly two metres (Panchen 1985). In contrast, Ossirarus was probably little more than 300 mm long, and much more similar in size to the late Viséan tetrapods from East Kirkton like the temnospondyl Balanerpeton (Milner and Sequeira 1994) and the stem amniotes Eldeeceon (Ruta et al. 2020) and Silvanerpeton (Ruta and Clack 2006). However, although the skull of Ossirarus was similar in size to those of the East Kirkton tetrapods, there is one notable difference: the orbits of Ossirarus are much smaller. This was an unexpected variation but may be explained by differences in ecology. The East Kirkton tetrapods are generally considered to be the earliest known example of a terrestrial fauna (e. g. Clack 2017) whereas the presence of lateral line canals in Ossirarus suggest it was either aquatic or amphibious and less reliant on vision for prey capture.

Axial skeleton.

The axial skeleton of Ossirarus is represented by a number of disarticulated cervical and trunk centra, neural arches and ribs (Fig. 7). The vertebrae are multipartite and consist of four parts: two central elements and a neural arch in bilateral halves. None is preserved intact and parts of numerous vertebrae are scattered on the left side of the specimen with most of the bones of the pectoral girdle and forelimbs on the right (Fig. 7).

Centra. The most anterior centrum lies c. 36 mm behind the postparietals and adjacent to the interclavicle (Fig. 7). It is preserved largely in ventral view but with part of the posterior edge and ‘ aperture’ for the notochord exposed. It is well preserved and appears to be uncrushed, but it is cracked along the ventral midline and one half has slightly overriden the other. The centrum forms a segment of a circle approximately 7 mm in diameter. It is c. 3 mm long, 2.5 mm high, and 0.5 mm thick. It would have surrounded a notochord c. 6 mm in diameter. The outer surface of the centrum is finished in periosteal bone. The centrum most closely resembles the atlas intercentrum of Acanthostega (Clack 1998, fig. 1) and given its position is most likely to be the atlas intercentum of Ossirarus.

Eleven cervical / trunk centra are preserved. The most complete are crescent-shaped in antero-posterior view and would have formed a thin husk of bone less than 1 mm thick around a notochord c. 6 mm in diameter (Fig. 8). In lateral view the centra are roughly triangular-shaped, c. 2.5 mm long, with the base the same length as the height of the sides. No facets are preserved for articulation with either the neural arch or the ribs, and there are no other features that may help distinguish the pleurocentra from the intercentra. Judging by the length of the neural arches, two centra would be accommodated beneath each arch, presumably with one occupying the position of the pleurocentrum the other the intercentrum. There is no evidence of paired pleurocentra typically found in rachitomous vertebrae or the dorsally fused pleurocentra of Whatcheeria (Lombard and Bolt 1995; Otoo et al. 2021). The centra of Ossirarus are probably the earliest known example of the gastrocentrous arrangement found in early tetrapods.

Parts of up to five neural arches are exposed (Fig. 7). They are preserved as bilateral halves, separated along the midline. They are approximately 6 mm long and the body of the neural arch is well ossified with short transverse processes projecting ventrolaterally from midway between the well-developed-zygapopheses (Fig. 8 C). In all cases the neural spine has broken off and have failed to be identified amongst the vertebral fragments. Judging by the position and length of the breaks, the neural spines occupied a posterior position and had a basal length of c. 2 mm. On the underside of each half of the neural arch is an area of unfinished bone, approximately square-shaped (Fig. 8 B), probably marking the area where the neural arch rested on the notochord. A similar scar is present on the neural arches of Eoherpeton, where the neural arch contacted the underlying pleurocentrum (Smithson 1985, fig 18).

Ribs. Some partial ribs are preserved (Fig. 7). Four at the anterior end on the scatter of post cranial bones are stout, straight rods c. 10 mm long, slightly expanded at their proximal ends but not obviously double-headed. There is no evidence of uncinate processes. Given their position behind the skull and beside the interclavicle, they are most probably cervical ribs. Immediately in front of the right humerus is the proximal end of a double-headed rib (Fig. 7). The rib head is clearly divided into dorsal tuberculum and ventral capitulum. The capitulum extends proximally beyond the tuberculum indicating the presence of a short transverse process on the corresponding vertebra (Fig. 8 C). Beside the left humerus is a short piece of rib shaft (Fig. 7). It is c. 8 mm long and c. 3 mm wide and the exposed surface is gently convex. It probably formed part of the shaft of a broad, flattened rib, of the type found in the pectoral region of Whatcheeria (Otoo et al. 2021, figs 2, 3).

Appendicular skeleton. The appendicular skeleton is represented by much of the dermal pectoral girdle, the left and right humeri and the left radius (Figs 7, 9 – 12). All the bones are disarticulated and displaced, the interclavicle and cleithra are broken, and most of the left clavicle, the anterior portion of the interclavicle and the entepicondyles of each humerus are missing and represented by faint impressions in the matrix. The interclavicle is preserved in internal (dorsal) view, the right clavicle is preserved in external (ventral) view.

Cleithrum. The cleithrum is a long, narrow bone, approximately 30 mm in length, with an expanded dorsal blade (Fig. 10 A). The right cleithrum is preserved in external view and broken into two pieces with the dorsal portion slightly overlying the ventral shaft. The posterior part of the dorsal blade is partially concealed by the left humerus. The left cleithrum is preserved in internal view and broken into three pieces slightly separated from one another. The dorsal blade is also partially concealed by the left humerus.

The cleithrum is divisible into two parts: a long narrow stem making up approximately two thirds of its length and an expanded dorsal blade. The stem is approximately 3 mm wide along most of its length but tapers slightly ventrally. In lateral view, it is gently bowed, with a convex posterior edge and a concave anterior edge. The internal surface carries a shallow central groove that fades out dorsally, where the stem expands to form the dorsal blade. The anterior edge is thin and sharp and may represent part of a post-branchial lamina (see Coates and Clack 1991; Coates 1996), but the posterior edge is gently rounded. Both diverge dorsally to produce the ventral portion of an expanded dorsal blade. The posterior edge of the blade is sinuous and terminates with a blunt dorsal process. In front of this process the dorsal edge is essentially straight and meets the anterior edge almost at right angles. It has a maximum anteroposterior length of c. 5 mm. The edges of the blade are gently rounded and the external surface of the right cleithrum is ornamented with a number of pits and grooves.

The cleithrum of Ossirarus is unlike that of most other early tetrapods in having a longer stem and a smaller and more angular dorsal blade. The cleithrum of the earliest known tetrapods Acanthostega, Ichthyostega and some specimens of Whatcheeria is co-ossified with the scapulocoracoid (Otoo et al. 2021), but in Pederpes (Clack and Finney 2005) and in one specimen of Whatcheeria (Otoo et al. 2021) the stem is broad and the dorsal blade more circular with a distinct notch on the posterior edge separating the blade from the stem. In Ossinodus (Warren and Turner 2004), the cleithrum is robust with distinct blade and stem, and it bears facets for articulation with the scapulocoracoid and clavicle. These facets are not developed on the cleithrum of Ossirarus. In colosteids like Greererpeton (Godfrey 1989), and in the baphetid Eucritta (Clack 2001), the cleithrum is gently curved and expands dorsally but lacks a distinct blade, while in the anthracosaur Proterogyrinus (Holmes 1984), the shaft is broad and straight with some slight widening dorsally.

Clavicle. The left clavicle is represented by impression of the ventral surface of the clavicular blade in the matrix in front of the right humerus (Fig. 7) together with a short piece of the clavicular stem. The right clavicle is partially exposed between the right cleithrum and left radius (Fig. 9). Much of the clavicular blade of the right clavicle is concealed beneath the right cleithrum and only the lateral part of the blade is visible. The base of the clavicular stem is preserved, but most of the stem is missing. The right clavicle is exposed in ventral view and the surface is ornamented with a well-developed reticulate pattern of ridges and grooves (Fig. 9 A) that is also faintly visible in the impression of the left clavicle. This pattern of ornament has been found in a number of early tetrapods, including Acanthostega (Coates 1996), colosteids like Greererpeton (Godfrey 1989), Doragnathus (Smithson and Clack 2013) and many temnospondyls (Holmes 2000). The base of the clavicular stem is broad with laminae projecting from both the anterior and posterior edges. Together, these probably formed an open tube along the clavicular stem that received the stem of the cleithrum. Judging by the shape of the anterior portion of the interclavicle and the impression of the left clavicle, the clavicular blade was shaped like a long triangle, similar to those of Doragnathus (Smithson and Clack 2013) and Greererpeton (Godfrey 1989) (Fig. 10).

Interclavicle. The interclavicle is preserved in dorsal view (Figs 9, 10). Much of the anterior portion is missing and represented by impression in the surface of the matrix, but most of the posterior portion is preserved. It is broken into a number of pieces which have separated slightly, although part of the left lateral edge is missing. The interclavicle is approximately diamond-shaped, slightly longer than wide and lacks a parasternal process. As reconstructed (Fig. 10 B), it is c. 36 mm long and c. 27 mm wide. Judging by the impression of the anterior portion, the ventral surface is ornamented with a reticulate pattern of ridges and grooves seen on the clavicle. The dorsal surface is smooth but not flat. Its contours are similar to those seen in the interclavicles attributed to Doragnathus (Smithson and Clack 2013). Extending laterally on either side from the centre of the interclavicle is a broad ridge. In Doragnathus this ridge corresponds with a groove on the ventral surface that accepts a ridge on the dorsal surface of the clavicular plate. Extending posteriorly from behind the centre of the interclavicle is a short midline ridge. This terminates at the posterior edge of the bone. A similar ridge is present on the interclavicles of Doragnathus as well as on an interclavicle described from Blue Beach, Nova Scotia (Anderson et al. 2015).

Humerus. Both left and right humeri are preserved (Figs 7, 11, 12). They are embedded in matrix and visible mainly in ventral view. The left humerus appears to have been flattened slightly, but the right is undistorted. The anterior edge and part of the posterior edge of the right humerus are exposed, and part of the dorsal surface was available for study after the bone was temporarily removed from the block. Much of the entepicondyle is missing in both humeri, but impression of the dorsal surface is preserved on the left. Each humerus is c. 17 mm long.

The humerus (Fig. 11) has the characteristic L-shape of early tetrapods. The proximal articulation is relatively broad and straight. Judging by the impression of the left entepicondyle this was well developed and square-shaped. The right humerus is twisted midway along the shaft and the angle of torsion is between 20–25 degrees. The insertions of the principal locomotory muscles from the shoulder to the proximal end of the humerus are clearly defined.

The proximal posterior edge is essentially straight and the pre-entepicondylar ridge is absent. The brachial foramen pierces the posterior edge of the humerus at the base of the entepicondyle, as it does in Mesanerpeton (Smithson and Clack 2018). The entrance of the foramen is not visible in dorsal view but it can be seen in ventral view (Fig. 11 A, B, F). The exit is through the posterior part of the ventral ridge (see below). It is slightly concealed by this ridge and does not form a distinct opening on the ventral surface of the entepicondyle. The insertion for the coracobrachialis muscle is marked by a furrow on the posterior half of the ventral surface of the humeral head.

The ectepicondyle appears to be quite prominent, but the distal part of it is buried in the matrix. In the right humerus, it is visible in posterior view (Fig. 11 A – C). It starts as a swelling level with the entrance of the brachial foramen and develops into a ridge that projects distally into the matrix. The latissimus dorsi process is borne on a low ridge which extends proximally from near the inception of the ectepicondyle.

The anterior edge of the right humerus is well preserved. The proximal end is marked by a fine ridge which extends distally from the articulating surface. There is no prepectoral space. The ridge swells to form the deltoid process on the anterodorsal surface and the pectoral process on the anteroventral surface. Beyond the pectoral process, the anterior edge bows dorsally and extends towards the radial condyle as a thin bony lamina, similar to that seen in Acanthostega (Smithson and Clack 2018, fig. 6). There is no distinct origin of the supinator muscle. The radial condyle is a relatively large, unfinished swelling on the anterodistal corner of the humerus, which is clearly visible in ventral and anterior views.

On the ventral surface, a ridge extends posterodistally from the distal edge of the pectoral process onto the entepicondyle. It consists of two parts, anteriorly forming the smooth distal slope of the pectoral process and posteriorly the thickened proximal edge of the entepicondyle, pieced by the brachial foramen. The ridge fades between these two parts, turning distally towards the radial condyle. Presumably, it represents the vestige of the ventral ridge of tetrapodomorph fishes like Tiktaalik (Shubin et al. 2006) and Gogonasus (Holland 2013).

Radius. The left radius is preserved beside the left humerus (Fig. 12 A, B). It is embedded in matrix and exposed in dorsomesial view. It is c. 10 mm long and approximately 60 % the length of the left humerus. This compares with a radius – humerus ratio of 62 % in Pederpes (Clack and Finney 2005), between 50 % and 60 % in Whatcheeria (Otoo et al. 2021, fig. 29 A – C), 46 % in Proterogyrinus (Holmes 1980, 1984) and 49 % in Baphetes (Milner and Lindsey 1998). The ratio in Acanthostega is c. 53 % (Coates 1996, fig. 15) and in Crassigyrinus is 72 % (Panchen 1985).

The radius is approximately square-shaped in section with each of the sides being of similar dimensions. The faces of the exposed ventral and mesial sides are gently concave, and they meet at a sharp ridge. The ventral surface is further excavated below the proximal articulation to form a short groove. There is no evidence of the ventral radial crest figured by Coates (1996, fig 17) on the radius of Acanthostega. The junction between the medial surface and the concealed dorsal surface also forms a ridge. A similar ridge is present in Archeria (Romer 1957) and Baphetes (Milner and Lindsey 1998), but it is absent on the radius of Crassigyrinus (Panchen 1985), Pederpes (Clack and Finney 2005) and Whatcheeria (Otoo et al. 2021). The proximal end of the radius of Ossirarus is gently rounded. The shaft tapers distally to an incompletely ossified or broken distal end. In its overall morphology the radius of Ossirarus is more like those of Greererpeton (Godfrey 1989) and Proterogyrinus (Holmes 1980, 1984), where the four sides of the shaft have similar proportions, than those of Acanthostega (Coates 1996), Baphetes (Milner and Lindsey 1998), Ossinodus (Warren and Turner 2004), Pederpes (Clack and Finney 2005) and Whatcheeria (Otoo et al. 2021), where the dorsal (extensor) and ventral (flexor) surfaces are much broader than the anterior (mesial of Milner and Lindsay 1998) and posterior (lateral) surfaces, giving the radius a flattened appearance.

Scales. Most of the fragmentary scales were removed during preparation, but one slightly damaged example is preserved in internal view on the left side of the block near the right humerus. It is approximately semi-circular, with a straight side and a gently curved side (Fig. 12 C, D). It is c. 2.5 mm long and c. 1.5 mm wide. The straight edge is thickened into a rounded ridge and the remainder of the scale is very thin apart from a narrow lip around the curved edge. In section the scale is gently curved with a concave internal surface and a convex external surface (Fig. 12 E), a form described by Clack and Milner (2015, p. 23) as comma-shaped.

Ventral scales (gastralia) have been described in various early terapods, including Acanthostega (Coates 1996), Crassigyrinus (Panchen 1985), Greererpeton (Godfrey 1989) Pederpes (Clack and Finney 2005) and Proterogyrinus (Holmes 1984). None is triangular-shaped, but each has a rounded ventral ridge along their axis and, apart from the spindle-shaped scales of Crassigyrinus, each is externally convex.