Erik Jarvik, who died on 11 January 1998, was one of the twentieth century's leading researchers on early vertebrates and one of the main characters of the ‘Stockholm school’ of vertebrate palaeontology. His contribution to our knowledge of Palaeozoic fish and amphibians is unsurpassed, even though his views on their evolutionary relationships were largely regarded as heterodox by the wider scientific community.

Jarvik was born on a farm near Utby, a village in southwestern Sweden, and first studied natural sciences and geography at Uppsala University. He turned to vertebrate palaeontology after an expedition to East Greenland in 1932 with Gunnar Säve-Söderbergh, and spent all his career at the Swedish Museum of Natural History in Stockholm. By 1923, the museum's palaeozoology department had become a world-famous centre for study of early vertebrates, thanks to Erik Stensiö, who had several foreign and Swedish disciples. Among them was Jarvik, who became a curator in the department in 1937 and was head of it from 1960 to 1972.

It has been said of Jarvik that he could be silent in seven languages. At meetings, his unusual theories on early vertebrate phylogeny, in particular the multiple origin of the four-legged land vertebrates (tetrapods), were often vividly opposed by colleagues. He rarely replied in public, but then, several years later, the response usually came in the form of a thoroughly argued paper, if not a huge monograph.

Jarvik liked to work in the large armchair in his Stockholm office, with a cup of Swedish boiled coffee at his side and a big cigar or a cigarette, painstakingly doing “his duty” — every day writing just half a page of some thoroughly elaborated text. This resulted in 51 papers, books and monographs, mostly on lobe-finned fish (a name he did not like, however) and tetrapods, but also on jawless vertebrates, basic vertebrate head structure, sharks and vertebrate origins. His book, Basic Structure and Evolution of Vertebrates (Academic, 1980, 1981), sums up his views and remains a highly valuable reference work.

Jarvik also took part in eight expeditions to Greenland and two to Spitsbergen between 1932 and 1969. In 1991, aged 84, he valiantly followed a field party at Miguasha, in Canada, where his pet fossil fish, Eusthenopteron foordi, had been discovered in Devonian rocks.

Jarvik's first important work was his doctoral thesis (1942) on the anatomy of the snout of the fossil lobe-finned (sarcopterygian) fish, which were already regarded as the possible ancestors of tetrapods. He noticed that, among these 370-400-million-year-old fossil fish, some (the porolepiforms) apparently shared the same snout structure with the urodeles (salamanders and newts), whereas others (the osteolepiforms) shared unique snout features with other tetrapods (frogs and amniotes). He thus suggested that tetrapods evolved twice, from two different fish groups, implying that limbs with digits appeared twice in vertebrate history. This polyphyletic theory of tetrapod origins was strongly opposed by most other palaeontologists. Nowadays, it is generally thought that tetrapods have a single origin; nonetheless, osteolepiform fish are certainly their close relatives.

For Jarvik, this work was decisive, because most of his later studies were aimed at supporting his theory. It led him to investigate the entire anatomy of fossil and Recent lobe-finned fish, and of the earliest known tetrapods that he collected in the 365-million-year-old rocks of East Greenland. His last work, published in 1996, was a monograph on the Devonian tetrapod Ichthyostega, one of the earliest known land vertebrates, which he liked to call the “four-legged fish” (a nickname introduced by a Danish cartoonist when Ichthyostega was first discovered in 1932 by Säve-Söderbergh). In the 1950s Jarvik's descriptions of this animal raised great interest among scientists and the public, because it was the first fossil evidence for an intermediate form between fish and tetrapods: it had limbs with digits but also a scaly fish tail with fin rays.

Jarvik's theory of the polyphyletic origin of tetrapods rests on his way of assessing character homology (that is, shared characters inherited from a common ancestor) when reconstructing the relationships between living and fossil groups. For Jarvik, two widely different characters in distant groups could be regarded as homologous as long as they belonged to the same segment of the ideal vertebrate head and shared the same embryonic derivation. Conversely, minor anatomical differences were sometimes regarded by him as evidence for widely different ‘bauplans’ (basic structures). Being either too broad or too narrow, this concept of homology often could not be used for phylogeny reconstruction. It is not surprising that such views, expressed at the dawn of evolutionary systematics in the 1940s or 1950s, were regarded as heretical.

In the late 1960s, with the advent of Willi Hennig's phylogenetic systematics (cladistics, which dethroned evolutionary systematics in the 1970s), Jarvik readily accepted Hennig's principles of phylogeny reconstruction and classification, because they embodied several ideas that had long been held by the Stockholm school, such as the rejection of primitive characters in phylogeny reconstruction, the unknowability of actual ancestors, the unreliability of the stratigraphical record, and the importance of embryonic development in assessing the primitive or advanced states of characters. Why Jarvik soon after rejected this major renewal in comparative biology remains obscure. It may be because he considered it was nothing new to him, or because he rejected the principle of parsimony (the cornerstone of cladistics), which threatened his polyphyletic theory of tetrapod origins.

Nevertheless, whether Jarvik was right or wrong about certain questions of vertebrate phylogeny is no longer important. What we now see as fundamental in his work is his constant care for the interpretation of fossils in the framework of the anatomy and embryology of their living relatives, based on observations and descriptions of outstanding quality.