The skull (as preserved, 162 mm in length, excluding the snout) consists of two fragments that comprise a significant part of the skull table and palate, along with the occipital region. A transverse section of the skull at the level of the anterior border of the orbits shows that the dermal bones are moderately thick (around 3 mm high). This value becomes significant, when compared with the less than 1 mm deep bones of the lydekkerinid Lydekkerina huxleyi (Jeannot et al., 2006). The skull table preserves the orbits and the postorbital area with the exception of both posterolateral corners, which were severely damaged by erosion. Moreover, most of the preorbital area is visible on the right side of the skull, where the almost complete right naris is present. The left external nostril may be concealed by sediment left in place. The tip of the snout is compressed to such an extent that the premaxillae are not preserved.

In ventral view, most of the palate is preserved, and only a small fragment on the left side (involving part of the palatine ramus of pterygoid, the left ectopterygoid and part of the palatine) is missing (Fig. 2), as are the posterolateral corners, which are strongly weathered.

The skull is approximately triangular in outline with slightly convex posterolateral corners, and it widens posteriorly at the level of the anterior margin of the orbits. The posterior margin of the table is smoothly concave, allowing the occipital condyles and part of the occiput to be visible in dorsal view. In lateral view, the skull table slopes gently anteriorly, and the occiput seems to have been distinctly higher than the snout, even considering the lateral compression undergone by the latter. In occipital view, the margins of both cheeks are smoothly convex and converge on the midline of the skull. Here, a pronounced concavity forms an extensive longitudinal trough, particularly evident at the level of the orbits. As a result, orbits are slightly raised above the table (Fig. 1A, B). The ventral margin of the skull is straight in occipital view, because the occipital condyles are at the same level as the quadrate condyles.

The orbits are relatively small, subcircular, and located at a roughly equivalent distance from the midline and from the lateral margins of the skull. Based upon our reconstruction, including the estimated length of the snout, orbits appear to have been situated immediately posterior to the skull table mid-length (Fig. 1C). A small pineal foramen is located in a subcircular depressed area, immediately anterior to the middle point of the suture between the parietals, well behind the posterior margin of the orbits.

The bones of the skull roof are covered by a coarse, mainly pitted ornamentation without pustules or conules at the junction of the ridges, unlike rhitydosteids in which pustules are visible at these junctions (Cosgriff and Zawiskie, 1979) and Uruyiella liminea, from the same unit (Piñeiro et al., 2007b); evident zones of intensive growth are also absent on the preserved areas. The ornamentation closely resembles that in rhinesuchids, particularly in the uniformity of the network of ridges enclosing mostly rounded depressions (CM pers. obs.). The lateral line sensory sulci are discontinuous and part of the supraorbital and infraorbital canals are clearly visible on the preserved skull. They are deeply impressed and curve smoothly to approach each other over the lacrimal. The infraorbital canal lacks an infraorbital flexure such as is observed in many Palaeozoic and Mesozoic temnospondyls (Yates and Warren, 2000), and extends posterolaterally producing a short portion over the jugal. Part of the interorbital section of the supraorbital sulcus is probably present at the right side, represented by a series of depressions over the prefrontal. Behind the orbits, the infraorbital sulcus is represented by short canals visible over the postorbitals and the left jugal (see Fig. 1). Despite its peculiar features, the described distribution of the sensory canals in Arachana appears to be similar to that present in rhinesuchids, and is different to that exhibited by Lydekkerina huxleyi and most lydekkerinids, where the infraorbital canal forms a step to Z-shaped flexure over the lacrimal (Hewison, 2007 and Jeannot et al., 2006). Nonetheless, it resembles the condition with the Australian lydekkerinid Chomatobatrachus halei (after Hewison, 2007, Text-figure 44; page 54).

Part of a narrow strip of the maxilla is present along the right anterolateral border of the skull, suturing with the lacrimal very close to the infraorbital sensory sulcus. Medially, it fails to contact the nasal as the lacrimal reaches the naris, in the preorbital area. This primitive condition is observed in forms such as Dvinosaurus, but it is also seen in the lydekkerinids Chomatobatrachus halei (Cosgriff, 1974 and Schoch and Milner, 2000) and Broomulus dutoiti (Hewison, 2007, Jeannot et al., 2006 and Shishkin et al., 1996) and is also present in older taxa such as the Carboniferous Iberospondylus (Laurin and Soler-Gijón, 2006). However, recent work by Warren et al. (2006) did not consider this character distinctive, and it was featured neither in the description nor in the Interpretative drawings of Chomatobatrachus. Jeannot et al. (2006) suggest a putative “intraspecific” variation of this character in Lydekkerina huxleyi, resulting in individuals that may exhibit different conditions, either the lacrimal entering the naris or a nasal-maxillary contact. The nares of Arachana are large, round, and placed close to midline of the snout.

The prefrontals are partially preserved on both sides of the skull, sufficiently to allow for a partial reconstruction. From the left preserved fragment, it is possible to place it on the anteromedial margin of the orbit. From the right side, its anterior margin can be followed approximately until the level of the middle length of the lacrimal.

Elongate frontals occupied the midline of the skull roof between the orbits. They are excluded from the orbital margin by a prefrontal-postfrontal contact. Posterolaterally, their suture with the post-frontals is almost parallel to the midline. Their contact with the nasals is not visible, but it is situated in front of the orbits. Anteriorly, they seem to project laterally where they reach their maximum width. This inference was based on the position of the preserved medial border of the prefrontal (Fig. 1).

The post-frontals are narrow, laterally curved bones that form the medial and posterior border of the orbit. They expand posteriorly to suture with the postorbital posterior to the orbits.

The parietals are distinctly trapezoidal bones, with their anterior margin being narrower than the posterior. The pineal foramen is located slightly forward of the midpoint of their length. As in most stereospondyls, the parietals suture anterolaterally with the post-frontals and laterally with the supratemporals; posteriorly they form a nearly straight contact with the postparietals.

The postorbitals are almost complete and border the orbit posterolaterally. They are relatively short and roughly trapezoidal, having their anterior border narrower than the posterior. A small portion of the postorbital sensory sulcus is present on the left postorbital, represented by a short canal, which apparently also enters the jugal.

The jugals are partially preserved on both sides; the left jugal shows its contribution to the lateral margin of the orbit and the shape and extension of its posterior portion, which appear to end anterior to the level of the parietal-postparietal contact. On the right side, the jugal extends well anterior to the orbit as is usual in rhinesuchids, lydekkerinids and other basal stereospondyls (Yates and Warren, 2000), but not in most rhytidosteids, in which the jugal does not extend far beyond the anterior level of the orbit or extends only slightly beyond this level.

Both supratemporals are incompletely preserved because of surface weathering. They appear to be roughly diamond-shaped bones, but it is not possible to determine if they contribute to the otic notch margin, as they do in rhinesuchids. We have reconstructed the skull as lacking that contribution because of the structure of this area in Arachana (e.g. shallowly incised otic notches).

Approximately quadrangular postparietals occupied the posteromedial, fairly concave border of the table. This concavity was apparently continuous with the medial border of the tabular horns, which are broken on both sides; their position can be determined by the shape and orientation of the paraoccipital processes. On both sides of the skull table, incomplete tabulars are present and part of the anterior margin of the right otic notch is preserved. This margin is located at the same level as the posterior border of the postparietals, and thus, the notch was not deeply incised along the margin of the table, as in the lydekkerinid Lydekkerina huxleyi (Hewison, 2007, Jeannot et al., 2006 and Shishkin et al., 1996).

The morphology of either the premaxillary region and the anterior palatal vacuity is unknown. The interpterygoid vacuities are large and elongated and slightly pointed anteriorly. They are transversely widest at the level of the palatine-ectopterygoid suture (Fig. 2) and are separated from each other by the slender cultriform process of the parasphenoid. The subtemporal fossae are partially preserved on both sides and they appear to have been teardrop-shaped. Only the right choana is preserved. This is anteroposteriorly elongate and located just in front of the anterior border of the vacuities.

The maxilla is a narrow strip of bone only partially preserved on the right and left sides of the palate and posterolaterally on the left margin of the skull. It forms the lateral margin of the choana. There is no contribution to this margin from an anterior extension of the palatine (Fig. 2 and Fig. 3). This condition is similar to that present in rhinesuchids, lydekkerinids and other basal temnospondyls (Shishkin et al., 1996, Watson, 1962 and Yates and Warren, 2000) and it differs from that of most mastodonsaurids (Damiani, 2001), rhytidosteids (Schoch and Milner, 2000 and Yates and Warren, 2000) and Chomatobatrachus halei (Warren et al., 2006), in which the contribution of the maxilla to the choana is reduced or minimal.

Along its length the maxilla forms straight sutures with the palatine and ectopterygoid, and it probably contributed to the formation of the anterolateral border of the subtemporal fenestra.

Both vomers are partially preserved along the anterior border of the interpterygoid vacuities. Although their surface was slightly damaged during preparation, it shows a smooth ornamentation including some patches of shagreen. Just in front of the choana, a replacement pit for the vomerine tusk can be seen on the right vomer. The medial margin of the choana bears a row of small teeth, which is continuous with a row along the posterior margin (Fig. 3). The vomers cover the anterior tip of the cultriform process with which they form an inverted V-shaped suture. The posterior extensions of the vomers do not overlap the lateral margins of the cultriform process unlike the condition in Lydekkerina huxleyi (processus parasphenoidales, sensu Jeannot et al., 2006). The vomers overlap the process along a short distance immediately anterior to the anterior border of the interpterygoid vacuities. The narrow and almost parallel-margined cultriform process is slightly convex and elliptical in cross section, without any signs of a ventral keel. Unlike in most rhinesuchids and lydekkerinids, except Deltacephalus whitei (Hewison, 1996 and Hewison, 2007), the cultriform process expands laterally its anterior portion and extends well anterior to the anterior margin of the interpterygoid vacuity.

Anterolaterally, on the right side of the skull, the palatine is complete and bears a large palatine tusk with its replacement pit, followed posteriorly by one or two teeth. The palatine-ectopterygoid contact is visible on the right side, immediately behind the most posterior palatine replacement pit for the palatine tusk. The posteromedial corner of the palatine is simple and does not extend posteriorly to the most anterior ectopterygoid tooth. This condition is similar to that present in all rhinesuchids. However, there is an elongated palatine posteromedial process in lydekkerinids, which extends posterior to the most anterior ectopterygoid tooth (Hewison, 2007, Jeannot et al., 2006 and Warren et al., 2006).

The long and narrow ectopterygoid stretches between the palatine ramus of the pterygoid and the maxilla. It bears a tooth row of at least 11 teeth, which are similar in size to or a little larger than the maxillary teeth. Ectopterygoid tusks are absent. We are not sure whether the ectopterygoid entered the subtemporal fenestra or whether an alar process of the jugal was present. Although a space is visible where the palatal exposure of the jugal normally occurs in other taxa, the broken pterygoid corpus precludes any firm conclusion. However, a posterolateral flange seems to have been absent from the posterior surface of the palatine ramus of pterygoid, bordering the subtemporal fossa.

The parasphenoid corpus is approximately hexagonal; its ventral surface is smoothly concave and weakly ornamented (Fig. 2 and Fig. 4). Laterally, it contacts with the pterygoids along a flat, posterolaterally directed inter-digitated suture, as in all stereospondyls (Yates and Warren, 2000). Posteriorly, the parasphenoid corpus contacts a small relict of an ossified basioccipital, which is visible in both ventral and occipital views of the skull. The exoccipitals occur just anterolateral to this contact (see Fig. 5 and Fig. 6). The presence of a small basioccipital between the exoccipitals characterizes rhinesuchids among stereospondyls (e.g. Schoch and Milner, 2000 and Watson, 1962). However, Watson (1919) suggested that the basioccipital was exposed between the exoccipital condyles in the lydekkerinid Lydekkerina huxleyi (see also Hewison, 2007, Text-fig. 21a) although, sutures were not observed in the corresponding area by Warren et al. (2006). Perhaps a thin ossified strip of this bone remains preserved in some lydekkerinid specimens (Shishkin, pers. comm. 2010), but detailed descriptions of them are not available to compare with the condition present in Arachana. The exoccipital condyles seem to have been located almost at level with or a little anterior to the quadrate condyles.

Posterolaterally, the parasphenoid corpus bears low crescent-shaped cristae musculari. Each crest surrounds ventromedially orientated muscular “pockets” which remain fairly separated (Fig. 2 and Fig. 4), as in all Palaeozoic temnospondyls and Triassic taxa such as lydekkerinids and mastodonsaurids (Damiani, 2001).

Both pterygoids are partially preserved and the left one is nearly complete. The pterygoid corpus is slightly dorso-ventrally concave and bears a faint ornamentation. Under magnification, sporadically distributed small patches of denticles can be observed (Fig. 2, Fig. 3 and Fig. 4), suggesting the presence of extensive shagreen. The palatine ramus extends anteriorly to make contact with the vomer at the level of the palatine tusk, where it narrows before disappearing beneath the vomer (Fig. 3). This condition excludes the palatines and ectopterygoids from the border of the interpterygoid vacuities, as in the rhinesuchid Rhineceps nyasaensis and other basal temnospondyls (see Laurin and Soler-Gijón, 2006, Watson, 1962 and Yates and Warren, 2000). Both quadrate rami are broken mesiodistally but they were nearly vertically orientated.

A shagreen of denticles, almost invariably damaged during preparation, is represented by scattered patches on the base and anterior one-third of the cultriform process, on the corpus and palatine ramus of the pterygoids, and on the vomers. Moreover, small, isolated denticles can be observed on the palatines and ectopterygoids, suggesting that shagreen was also developed on these bones. Widespread shagreen of denticles on the palatal series is present in rhinesuchids and rhytidosteids, but it is observed just in the parasphenoid and pterygoids of Lydekkerina huxleyi (e.g. Jeannot et al., 2006 and Warren and Black, 1985; CM pers. obs.).

A continuous tooth row is visible on the vomers (medially and posterior to the choana) and this continues on the palatines, and ectopterygoids. The row includes tusks and teeth of variable sizes that are of equivalent size to or slightly larger than those preserved on the maxilla. All marginal teeth curve lingually, and are anteroposteriorly compressed; they are, thus, oval in transverse cross-section. In contrast, the palatine tusks and their replacement pits are comparatively larger than the marginal teeth, and their bases are circular. Well-developed carina can be observed on the external surface of all the teeth.

The palatine tusk is followed by just one or probably two teeth, a feature that differentiates Arachana from both rhinesuchids and lydekkerinids except L. huxleyi, where a well-developed tooth row is present (Watson, 1962). The condition in Arachana is more similar to some lydekkerinids, which possess a small number of teeth behind the palatine tusk (Jeannot et al., 2006 and Shishkin et al., 1996). Due to the great asymmetry and variation recorded in temnospondyl taxa, comparisons between different types of dentitions cannot be assessed with great confidence (Watson, 1962).

A large replacement pit of the vomerine tusk is preserved anterior to the anterior margin of the choana. The presence of vomerine tusks has been described for lydekkerinids, as well as in most other temnospondyls, including rhinesuchids (CM pers. obs.).

The central part of the occiput is dominated by exoccipitals. They contact the postparietals dorsally, and extend dorsolaterally onto the paroccipital process where they meet the ventral extension of the tabular. The paroccipital process is robust and rod-like, having an irregular surface near its contact with the parotic process of the tabular, where a round and relatively large vagal foramen is located. The paroccipital processes are dorsolaterally orientated and form the ventral margin of the posttemporal fenestra. The latter is relatively large and teardrop-shaped as in most stereospondyls excluding rhinesuchids, in which the fenestra is very thin and laterally expanded due to the almost horizontal orientation of the paroccipital process (Watson, 1962; CM pers. obs.). Medially, the exoccipitals enclose the foramen magnum, and ventrally, they are sutured with a small ossified basioccipital, which forms a narrow strip of bone between the exoccipital condyles (Fig. 5 and Fig. 6). The latter are formed entirely by the exoccipitals and the condylar facets are nearly flat and face somewhat medially. The occipital condyles are visible in dorsal view, and positioned at level of the quadrate condyles.

A tall, posteriorly convex ascending ramus of the pterygoid arises from the occipital border of the quadrate ramus and broadly contacts the squamosal. On the lower margin of the quadrate ramus of the pterygoid is an apparently shallow and elongated ledge. It is partially preserved because of the damage of the specimen in this area. This ledge is similar to that described for Lydekkerina huxleyi and Eolydekkerina magna as the crista obliqua pterygoidei (Jeannot et al., 2006 and Shishkin et al., 1996), and apparently differs from the tall and sharp-edged crista obliqua present in rhinesuchids (see Watson, 1962).

Small triangular posterior projections are present lateral to the parasphenoid-pterygoid suture, which can also be seen in a ventral view of the skull (Fig. 4 and Fig. 5). These projections were previously described in Lydekkerina and Eolydekkerina as “lappets” belonging to the parasphenoid and providing attachment for the stapes (Watson, 1919). Shishkin et al. (1996) suggested that they are formed by the contribution of both the parasphenoid and the pterygoid. In Arachana, the triangular projections seem to belong to the pterygoid, as they are present immediately lateral to the pterygoid-parasphenoid suture; the posterior border where these bones meet each other forms a slightly excavated edge. The transverse notch characteristic of lydekkerinids (sensu Shishkin et al., 1996) separating the lappet from the subotic procces of the exoccipital, is not observed in Arachana. The identity of both previously described structures has been recently questioned by Jeannot et al. (2006) who argued that the “lappets” represent the most extreme lateral projection of the muscular cristae and the notch is produced as an effect of the exoccipital recess behind the pockets. In Arachana, the posterior projections of the pterygoid do not appear to be related to the crista muscularis and even with the presence of an exoccipital recess behind the pockets, a parasphenoid notch seems to be absent.

Both left and right stapes are preserved in anatomical position (Fig. 4, Fig. 5 and Fig. 6). They are relatively robust, slightly anteroposteriorly compressed, and rest in an apparently firm connection against the parasphenoid and exoccipital lateral margin. A similar condition is also present in many Palaeozoic temnospondyls and the South African lydekkerinids (e.g. Jeannot et al., 2006 and Yates and Warren, 2000). The massive footplate bears a small stapedial foramen facing posteromedially, close to the base of the shaft. The stapedial shaft is also robust and of nearly equivalent size to the paroccipital process; it curves slightly dorsolaterally to end distally just underneath the otic notch. The shaft bears an oblique ridge (sensu Bystrow and Efremov, 1940) along its ventral surface, which has not been described in lydekkerinids or in rhinesuchids.