The Middle Cambrian Pikaia gracilens (Walcott) has an iconic position as a Cambrian chordate, but until now no detailed description has been available. Here on the basis of the 114 available specimens we review its anatomy, confirm its place in the chordates and explore with varying degrees of confidence its relationships to both extant and extinct chordates and other deuterostomes. The body of Pikaia is fusiform, laterally compressed and possesses about 100 myomeres. The head is small, bilobed and bears two narrow tentacles. There is no evidence for eyes. Apart from a thin dorsal fin (without finrays) and a series of at least nine bilaterally arranged appendages with possible pharyngeal pores at the anterior end, there are no other external features. In addition to the musculature the internal anatomy includes an alimentary canal, the anterior of which forms a prominent lenticular unit that is almost invariably preserved in positive relief. The cavity is interpreted as pharyngeal, implying that the mouth itself was almost terminal. The posterior extension of the gut is unclear although the anus appears to have been terminal. The most prominent internal structure is a reflectively preserved unit, possibly hollow, termed here the dorsal organ. Although formerly interpreted as a notochord its position and size make this less likely. Its original function remains uncertain, but it could have formed a storage organ. Ventral to the dorsal organ a narrower strand of tissue is interpreted as representing the nerve chord and notochord. In addition to these structures, there is also evidence for a vascular system, including a ventral blood vessel.
The position of Pikaia in the chordates is largely based on the presence of sigmoidal myomeres, and the more tentative identification of a notochord. In many other respects, Pikaia differs from the expected nature of primitive chordates, especially as revealed in amphioxus and the Cambrian record (including Cathaymyrus, Haikouichthys, Metaspriggina, Myllokunmingia, and Zhongxiniscus). Whilst the possibility that Pikaia is simply convergent on the chordates cannot be dismissed, we prefer to build a scenario that regards Pikaia as the most stem-ward of the chordates with links to the phylogenetically controversial yunnanozoans. This hypothesis has implications for the evolution of the myomeres, notochord and gills. Finally, the wealth of material of Pikaia indicates that, although by definition there must be some sort of taphonomic imprint, the consistency of preservational details allows a reliable reconstruction of the anatomy and does not significantly erode phylogenetically relevant characters.