SUMMARY1. The enzyme horseradish peroxidase (HRP) was injected into single axons that innervated hair follicle receptors to study the morphology of their collaterals in the dorsal horn of the cord. The axons were impaled near the dorsal root entrance zone in the lumbosacral spinal cord of anaesthetized cats and HRP injected by passing current through the intra-axonal micro-electrode. The morphology was revealed by subsequent histochemistry.2. Thirteen hair-follicle afferent fibres were stained including six that innervated tylotrichs (type T hair follicle afferent units) and one that innervated guard hairs (type G unit). The remaining six axons were not classified according to hair type, but, on the basis of their axonal conduction velocities, would have been either type G or T.3. Eleven axons could be traced back into the dorsal roots. Eight of these, upon entering the cord, turned and ran towards the brain. They did not divide into rostral and caudal branches. Three of the eleven did divide and gave rise to both rostral and caudal branches.4. Sixty-three collaterals were given off the thirteen stained axons. All well-filled collaterals had a strikingly similar morphology. They descended through laminae I-III of the dorsal horn into the deeper parts of lamina IV or into lamina V, before turning and ascending back into superficial lamina IV and lamina III where they branched profusely to give rise to their terminal arborizations. Terminal boutons, most commonly of the 'en passant' type, were numerous in lamina III, but were also seen in the dorsal part of lamina IV and in ventral lamina II. None were observed in

Intracellular injection and subsequent histochemical localization of horseradish peroxidase have been used to stain the soma, dendrites, axons, and axon collaterals of spinalcervical tract neurons and unidentified dorsal horn neurons in the cat. This technique may be used in combination with the intracellular injection of Procion yellow to demonstrate by light microscopy connections between physiologically typed vertebrate neurons.

SUMMARY1. The morphology of physiologically identified spinocervical tract neurones was studied using the intracellular injection of horseradish peroxidase in anaesthetized cats.2. Thirty-six spinocervical tract neurones were reconstructed from serial sections of the lumbosacral spinal cord, cut in either the transverse or longitudinal planes.3. Horseradish peroxidase provided a more complete picture of the dendrites of spinocervical tract neurones than earlier experiments using Procion Yellow injection (Brown, House, Rose & Snow, 1976a). The longitudinal (rostro-caudal) spread of dendrites from an individual cell was much greater in the present material; neurones in the medial parts of the dorsal horn had dendrites extending for about 500,um from the soma (1 mm total spread) and neurones in the lateral horn had dendrites extending for about 1 mm from the soma (2 mm total spread). However, the conclusions of the earlier work, on the medio-lateral and dorso-ventral extents of dendritic trees, together with the shapes of dendritic trees viewed as reconstructions in the transverse plane, have been confirmed. Dendrites of spinocervical tract cells barely entered lamina II of Rexed: they often ran in the longitudinal direction along the border between laminae II and III for several hundred asm. Dendritic spines were observed on many spinocervical tract neurones.4. Horseradish peroxidase reaction product stained up to 2-5 cm of the axon of spinocervical tract neurones. Axons usually pursued a

1. In acute experiments, detailed grids of micro-electrode recordings were made from spinocervical tract (s.c.t.) cells in the lumbosacral cord of anaesthetized cats. These grids provided electrophysiological data on the location, distribution, density and somatotopic organization of s.c.t. neurones.2. In acute experiments lasting up to 48 hr, retrograde labelling of s.c.t. cells was carried out by injecting horseradish peroxidase into the lateral cervical nucleus in anaesthetized cats. The aim was to flood the nucleus with the enzyme so that all s.c.t. neurones would be labelled in order to provide an independent check on the location, density and distribution data obtained in the electrophysiological experiments.3. The electrophysiological and the anatomical experiments were sometimes performed on the same animal.4. The electrophysiological and anatomical results were in excellent agreement. (a) S.c.t. cells are located mainly in ipsilateral laminae III, IV and V. About 25% are in lamina III, 60% in lamina IV and 10% in lamina V. There are a few cells in laminae I, II and VI making up the remaining 5%. (b) There are about 550-800 s.c.t. cells in the lumbosacral enlargement (L4-S2 inclusive) on each side of the cord. Most cells are in L7-S1 where there are twenty to forty s.c.t. neurones in each millimetre length of cord.5. Many marginal (lamina I) cells were labelled with the retrograde horseradish peroxidase method and a few cells on the contralateral side in laminae III-V and VII-VIII were also labelled. The marginal cells formed 12-18% of labelled cells on the side of injection and, in addition, a similar absolute number of marginal cells was labelled on the side contralateral to the injection. The possible identity of these neurones is discussed.6. S.c.t. cells form a sheet of neurones across the dorsal horn. The sheet is organized somatotopically in a way which resembles the representation of the hind limb dermatomes in the dorsal columns (Werner & Whitsel, 1967). There is a relative enlargement of the L6-S1 dermatomes which encroach upon their neighbouring cord segments. The gradient of the map is very steep in the mediolateral direction but gradual in the rostrocaudal direction. The detailed somatotopic maps generated in the present work have revealed that s.c.t. cells are arranged so that their receptive fields form longitudinal columns and cells within the columns have overlapping fields.

1. The neck muscle biventer cervicis is supplied by five separate nerve bundles that originate from segments C2-C5 and enter the muscle at different rostrocaudal levels. We have used the glycogen-depletion method to investigate the distribution of muscle fibers supplied by each nerve bundle and also the extent of motor-unit territories supplied by single motoneurons in the C3 segment. 2. Prolonged intermittent stimulation of each nerve bundle produced glycogen depletion in a compartment of muscle fibers that ran only a fraction of the whole-muscle length. The depleted compartment was separated by tendinous inscriptions from adjacent, serially arranged compartments that were supplied by different nerve bundles. Thus the muscle was divided into five in-series compartments, arranged in the same rostrocaudal sequence as the nerves by which they were supplied. 3. Six fast, glycolytic (FG) and five fast, oxidative-glycolytic (FOG) motor units were depleted by repetitive intracellular stimulation of their antidromically identified motoneurons in the C3 segment. The fibers of each motor unit were confined to a striplike subvolume whose cross-sectional area was only 20-40% of that for the whole compartment in which it was located. Single motor units contained an average of 408 extrafusal fibers (range: 262-582 fibers), and these were distributed with an average density of 20 fibers/mm2 in cross sections through their motor domains. No significant differences were found between the numbers or densities of fibers in FG and FOG motor units. 4. The specialized in-series organization of compartments has functional implications because the forces generated by one compartment of motor units must be transmitted through other in-series compartments of muscle fibers rather than directly onto skeletal attachments. The confined distribution of muscle fibers belonging to a single motor unit suggests that an additional level of organization may exist within individual compartments. The implications of these features for the physiological behavior and neural control of biventer cervicis are discussed.