The formic acid preparation of a nodule collected from the Lower Triassic of Spitsbergen, and containing a skull with some associated postcranial elements of Lyrocephaliscus euri (WIMAN), precises and enriches previous descriptions of the reputedly well-known trematosaur. lt permits more detailed descriptions of the occipital region, of the course of the carotid arteries and collateral vessels, and refutes the previous description of a stapedial foramen in trematosaurs. The discovery of pleurocentres associated with the other vertebral components confirms the primitive rhachitomy of trematosaurs. 

The anatomy of Norselaspis glacialis n.g., n.sp., a primitive kiaeraspidian from the Lower Devonian of Spitsbergen, is described on the basis of spécimens studied by grinding sections or prepared with dilute formic acid. This study yielded some new anatomical details, including the presence of a canal prolonging posteromedially the canal alloted to the facial nerve by Stensiö. This posterior prolongation of the « facial canal ›› into the posterolateral part of the labyrinth cavity is consistent with the hypothesis put forward by Allis, Lindström, Jefferies and Whiting, that this canal housed the glossopharyngeus nerve. Furthermore, in N. glacialis, the foramen usually referred to as the foramen for the œsophagus opens posteriorly into a cavity in the postbranchial wall, referred to here as the intramural cavity, and which is interpreted as having housed the heart. Consequently, the œsophagus probably accompanied the dorsal aorta through the aortic canal. Finally, the foramen generally interpreted as having transmitted the ventral afferent arterial trunk is here considered as having housed the hepatic vein, which emptied into the venous sinus of the heart. The ventral afferent arterial trunk may thus have passed through the former «œsophageal ›› foramen.
The problem of the position of the dorsal nerves in the Osteostraci is discussed, and it is suggested that the three foremost nerve canals opening into the oralobranchial cavity housed the maxillary ramus of the trigeminus, the facial nerve and the glossopharyngeus nerve respectively. The mandibular ramus of the trigeminus must have accompanied one of the two foremost nerves, but for the moment it is impossible to decide which.
The problem of the nature of the interbranchial crests of the Osteostraci is briefly discussed. Comparison with the branchial apparatus of the Petromyzontida does not support the hypothesis that the interbranchial crests are part of the branchial arches, incorporated into the endoskeletal shield. A different hypothesis is proposed, that the branchial skeleton of the Osteostraci was situated entirely inside the oralobranchial cavity, and was attached to the endoskeletal shield only by the ventromedial processes. The grooves classically allotted to the efferent branchial arteries would thus have housed extrabranchial arteries, branching off from the dorsal aorta, and irrigating the ventral branchial musculature.
A phylogeny and a classification of the kiaeraspidians are proposed. The evolution of this monophyletic group is characterized by, e.g., reduction of cornual processes, shortening of the abdominal division of the shield, subdivision of the lateral fields, and enlargement of the supraoral fossae.
The phylogenetic position of the kiaeraspidians within the Osteostraci remains uncertain. Their sister-group may be either the benneviaspidiens or the thyestidians, or Thyestes alone (in which case they would have to be included within the thyestidians).