Evidence is given that the median and lateral longitudinal giant myelinated fibres in the central nervous system of the prawn Leander serratus are syncitial structures, each formed by the fusion of the processes of many segmental nerve cells. Septa are found at intervals in the axoplasm of the median fibres, but they never completely transect it. They are probably relics of a condition similar to that in the earthworm where the giant fibre running the length of the cord is formed of a chain of segmental syncitial axons each divided from its neighbour by a complete septum which presumably functions as a synapse. The motor giant fibres, which are segmental and pass out of the central nervous system to the muscles, are the processes of single cells: the axoplasms of the two fibres of the pair in each segment undergo complete fusion with each other and then redivision before leaving the central nervous system. These motor giant fibres are non-myelinated within the central nervous system, although as great in diameter as other heavily myelinated fibres. They are myelinated outside the central nervous system. In the prawn therefore myelin sheath thickness is not an invariable function of axon diameter. The lateral giant-fibre synapses show complete axoplasmic discontinuity and their structure does not support Johnson's creation of a new category of synaptic relations. Two types of synapses between fibres are described. In the first, found in the lateral giant-fibre chain, two myelinated fibres lie closely side by side for a considerable distance, but their neuroplasms are separated by a myelin layer except over an extent of less than 10<latex>$ \mu $</latex>. In the second type, found at the point of contact of both the median and lateral fibres with the motor fibres, a myelinated fibre has synaptic connexions with a large non-myelinated fibre through many fine axonic processes which pass out through a small gap in the myelin sheath.