[1] 10w Merent Connections of the Oculomotor Nucleus i n the Chick

[1] 219w Horseradish peroxidase was injected into the oculomotor nucleus of the chick in order to locate and characterize the neurons projecting to this nucleus. In the rostral mesencephalon, 120-180 neurons were labelled in the medial area of the ipsilateral nucleus campi Foreli; 190-220 in the interstitial nucleus of Cajal (most of them contralateral); and smaller numbers bilaterally in the medial mesencephalic reticular formation, the nucleus of the basal optic root complex, and the central grey matter. More caudally, numerous neurons were labelled in the contralateral abducens nucleus and the vestibular complex and a few in the nucleus reticularis pontis caudalis. Labelled neurons appeared ipsilaterally in the caudal region of the nucleus vestibularis superior and in the rostral tip of the nucleus descendens just lateral to the tractus lamino-olivaris. In the contralateral vestibular complex, a group of labelled cells observed in the dorsolateral area may be homologous to the mammalian cell group Y. At the level of the contralateral abducens nucleus, the most numerous group of cells (625-700) projecting to the oculomotor nucleus formed a lateromedial fringe that affected the nucleus tangentialis, the rostral tip of the nucleus descendens, and the ventrolateral region of the nucleus medialis. Only a few labelled neurons were seen in the contralateral nucleus vestibularis superior, the ipsilateral cell group A, and the ipsilateral nucleus vestibularis medialis.

[1] 115w HRP injections were located lateral to the oculomotor nucleus (OCM) without affecting the fasciculus longitudinalis medialis (FLM), lateral to the oculomotor nucleus (affecting both the FLM and, to a lesser extent, the OCM), straddling the midline (with both oculomotor nuclei affected), centered in the OCM (but extending beyond its boundaries), and centered in the somatic component of the OCM, with a central dense core, presumably the region of effective uptake (Mesulam,'82), wholly contained within the boundaries of this component. The description that follows is based mainly on the results of the three injections of the latter type. Comparison with the results of injections in the other four locations were also used to clarify certain questions.

[2] 50w In the rostral mesencephalon (Fig. l), a group of 120--180 almost exclusively ipsilateral labelled neurons were located near the midline in the medial region of the nucleus campi Foreli (CF). These stretched caudally for 1,000 to 1,300 pm from a point some 1,800 to 2,000 pm rostral to the anterior

[3] 365w A AB AL An Aq, At B BCD BOR Cb CF CP CT CTz D Da Dd Dv EW FLM FRM GCt IP IS L LE LF Li LO LS M Mc Md Mvl MV nBOR NIII NVI NVIII nVII OCM OM 0s PL PM QF R RP RPgc Ru RxVM S SCE SCI SOP T TPc TrO V a Abbreviations cell group A nucleus nervi abducentis ansa lenticularis nucleus angularis aqueductus cerebri cell group B brachium coNunctivum descendens basal optic root cerebellum nucleus campi Foreli commisura posterior commisura tectalis corpus trapezoideum nucleus vestibularis descendens nucleus of Darkschewitsch nucleus Deiters dorsalis nucleus Deiters ventralis nucleus of Edinger-Westphal fasciculus longitudinalis medialis formatio reticularis medialis mesencephali substantia grisea centralis nucleus interpeduncularis nucleus interstitialis of Cajal nucleus laminaris left labelled fibers lingula tractus lamino-olivaris lemniscus spinalis nucleus vestibularis medialis nucleus magnocellularis nucleus vestibularis medialis (dorsomedial part) nucleus vestibularis medialis (ventrolateral part) nucleus motorius nervi trigemini nucleus of the basal optic root complex nervus oculomotorius nervus abducentis nervus octavus nucleus nervi facialis nucleus nervi oculomotorii tractus occipitomesencephalicus nucleus olivaris superior nucleus pontis lateralis nucleus pontis medialis tractus quintofrontalis nuclei raphes nucleus reticularis pontis caudalis nucleus reticularis pontis caudalis, pars gigantocellularis nucleus ruber radix mesencephalicus nervi trigemini nucleus vestibularis superior stratum cellulare externum stratum cellulare internum stratum opticum nucleus tangentialis nucleus tegmenti pedunculo-pontinus, pars cornpacta tractus opticus ventriculus tip of the oculomotor nucleus. These cells were multipolar (ff = 0.71 k 0.08) and had medium-sized perikaryon (Dc = 18.4 * 2.9 pm; Dmax = 21.5 * 4.9 pm), and some of their labelled axons were seen to course toward the fasciculus longitudinalis medialis (FLM) until they entered this fasciculus a t its rostral end (Fig. 2A,B These coincided over 75-225 pm with the caudal tip of the above group and stretched back caudally to the level of the rostral 225-375 pm of the oculomotor nucleus. Most of these neurons lay among the bundles of the fasciculus longitudinalis medialis or immediately ventral or dorsal to them (Fig. 2C). A few labelled cells were found bilaterally in the central grey matter (GCt), both at the level of the interstitial nucleus of Cajal and, more caudally, at the level of the oculomotor nucleus.

[4] 134w A much less numerous set of labelled neurons was sparsely distributed throughout the rostrocaudal extension of the nucleus of the basal optic root complex (nBOR). Some 30-45 of these neurons were found ipsilaterally (most in the dorsal part of the nucleus) (ff = 0.77 * 0.07; Dc = 18.3 t 2.8 pm; Dmax = 20.5 t 3.3 pm) and 25-30 in more central positions of the contralateral nucleus (ff = 0.83 t 0.05; Dc = 20.9 f 3.0 pm; Dmax = 22.8 * 4.3 pm). A few small labelled cells (ff = 0.73 * 0.06; Dc = 13.3 t 1.7 pm; Dmax = 15.6 t 2.9 pm) were found near labelled fibres seen to course dorsomedially from the nucleus of the basal optic root complex toward the fasciculus longitudinalis medialis and the oculomotor nucleus.

[5] 113w Labelled neurons were also found scattered bilaterally in the medial mesencephalic reticular formation near the interstitial nucleus of Cajal, the oculomotor nucleus, and the oculomotor roots (Fig. 2D). These neurons were most numerous in those animals in which the injection site was largest or in which the HRP deposit in the ventral oculomotor subnucleus was abundant. When small injections whose area of presumed effective uptake was limited to the somatic oculomotor nucleus were made, only a very few neurons were labelled in the contralateral pretectal area; when the injections were larger or more lateral, so that the Edinger-Westphal nucleus or the posterior commissure was affected, the number of labelled pretectal neurons was larger.

[6] 58w Finally, scattered labelled neurons also appeared in some animals in other mesencephalic areas such as the formatio reticularis lateralis, the Darkschewitsch nucleus, or the deep layers of the optic tectum. However, these scattered neurons were most numerous in animals with the largest injection sites and in those in which HRP had been injected just outside the oculomotor nucleus.

[7] 115w Contralateral to the injection site, 325-375 labelled neurons were found in the abducens nucleus (AB), most in the rostral two-thirds (Figs. 3, 4). At the rostral tip of the nucleus, these cells lay in a dorsal position near the floor of the IV ventricle. More caudally, they tended to surround the motoneurons (occupying mainly dorsal and lateral areas), although some were also found mingled with the motoneurons in the centre of the nucleus. Some 70-80% of the labelled neurons were small and elongated; the rest had perikarya similar in size and shape to those of the motoneurons. Further details of these internuclear neurons and the abducens nucleus have been published elsewhere (Labandeira-Garcia et al., '87).

[8] 98w A few labelled neurons were found bilaterally in the nucleus reticularis pontis caudalis (RP), just lateral to the roots of the abducens nerve. The great majority lay at the level of the caudal third of the abducens nucleus. When the injection was limited to the oculomotor nucleus, only isolated neurons were seen in other areas of the formatio reticularis pontina, but when the injection site was larger or lateral to the fasciculus longitudinalis medialis, labelled neurons appeared in the nucleus reticularis pontis caudalis in positions medial to the roots of the abducens nerve and in the nucleus paramedianus.

[9] 220w Following Wold ('76), the vestibular complex of the chick was considered to be composed of the following nuclei or cell groups: the nucleus vestibularis superior (S), cell group A (A), the nucleus Deiters dorsalis (Dd), the nucleus tangentialis (T), cell group B (B), the nucleus vestibularis medialis (M), and the nucleus vestibularis descendens (D). Most of the numerous vestibular complex neurons labelled by injection of HRP into the oculomotor nucleus appeared in sections that also intersected the abducens nucleus (Fig. 3). The most rostral group was located in the ipsilateral nucleus vestibularis superior. This group extended 1,125 to 1,275 p m , from a point 675-825 pm rostral to the anterior tip of the abducens nucleus, to end with the rostral third of this nucleus. About 60% of the 220-250 neurons comprising this group were found in the 300450 pm rostrocaudal stretch centered at the level of the rostral tip of the abducens nucleus, where they formed a tight, rounded cluster near the centre of the nucleus vestibularis superior (Fig. 5B). More rostrally, labelled cells were sparser and tended to ventromedial locations. Only 15-30 labelled neurons were observed in the contralateral nucleus vestibularis superior. The perikarya of labelled nucleus vestibularis superior neurons had ff = 0.82 * 0.5, Dc = 22.8 t 2.6 pm and Dmax = 24.6 4.1 wm.

[10] 405w Contralateral to the oculomotor nucleus treated with HRP, a compact group of 320-400 cells were labelled in a dorsolateral position. These cells stretched back from a level 225-375 pm rostral to that of the anterior tip of the abducens nucleus, to end with the rostral third of this nucleus. They therefore coincided in the same sections as the greatest density of labelled ipsilateral nucleus vestibularis superior neurons, which they were individually somewhat larger t h a n (ff = 0.81 f 0.06; Dc = 25.7 * 3.2 pm; Dmax = 27.9 k 4.6 pm). This lateromedially elongated group lay dorsal and roughly perpendicular to the pars dorsalis tracti spinocerebellaris (Fig. 5A); more caudally (Fig. 5C), it was just dorsal to the rostral 300-375 pm of the nucleus cochlearis angularis (An). No labelled neurons were observed in this location on the same side as a unilateral injection. 75-pm sections taken every 225 pm from a single representative animal and are numbered rostrocaudally. The largest group of labelled neurons, some 625-700, formed a contralateral horizontal band stretching from the point at which the fibres of the vestibular nerve entered the brain stem to the dorsomedial region of the nucleus vestibularis medialis (Fig. 4). This band thus affected the territory of the nucleus tangentialis, the rostral levels of the nucleus vestibularis descendens, and the ventrolateral part of the nucleus vestibularis medialis, roughly coinciding rostrocaudally with the motoneurons of the abducens nucleus. Only some 15-20% of t h i s group (ff = 0.77 k 0.06; Dc = 21.4 c 3.2 pm; Dmax = 22.7 c 4.3 pm) appeared in sections through the rostral third of the abducens nucleus, most in quite lateral positions corresponding to the nucleus tangentialis (Figs. 3, 5E). About 60% lay at the level of the middle third of the abducens nucleus. The more rostral of this 60% were mostly located lateral to the lamino-olivary tract (LO), whereas in the more caudal sections there were fewer labelled neurons lateral to the lamino-olivary tract, and a large number of small cells (ff = 0.78 2 0.06; Dc = 17.5 k 1.7 pm; Dmax = 21.7 -t 3.7 pm) appeared medial to it. At the level of the caudal third of the abducens nucleus, 90 % of the contralateral labelled vestibular neurons in the horizontal band were medial to the lamino-olivary tract. In the dorsomedial part of the nucleus vestibularis medialis, few isolated labelled neurons were observed.

[11] 75w In addition to the nucleus vestibularis superior neurons mentioned above, smaller numbers of ipsilateral labelled neurons appeared elsewhere: a compact group of 120-130 (ff = 0.80 k 0.07; Dc = 24.8 c 2.5 pm; Dmax = 27.8 * 4.5 pm) in a region just lateral to the lamino-olivary tract a t the level of the caudal half of the abducens nucleus (80-90 % at the level of its caudal third) (Fig. 5D); 10-20 neurons scat-

[1] 121w The vertebrate visual system appears to have evolved mainly in two directions: toward the acquisition of a wide visual field and panoramic vision or toward frontal vision with convergence and binocular overlap. Extreme examples of either tendency are provided by the frontal binocular vision of the primates and the lateral vision of the chamaleon, with its highly uncoordinated eye movements (Walls,'62). Various bird species have been shown (Bloch et al., '84; Jahnke, '84; Martin,'86a,b) to possess both visual modes, with frontal binocular vision generally used for close observation of nearby objects and lateral monocular vision used to detect distant moving objects. The relative importance of the two modes depends on the habits of each particular species, especially on its feeding habits.

[2] 83w For many years it has been thought (Kare and Rogers, '76) that the binocular part of the avian visual field was determined mainly by the shape of the head, the shape of the eye, and the position of the eye in the head: birds with flat, lateral eyes, such as pigeons and chicks, were considered to have smaller binocular fields than birds with tubular frontal eyes, such as owls and hawks. In spite of the existence of well-developed extraocular muscles, it was assumed

[3] 289w In HRP studies, it is of the utmost importance to avoid unwanted labelling due to enzyme uptake by tissues adjacent to the structure of interest. In the case of the chick oculomotor nucleus, avoiding such artifacts is made easier by the fact that the OCM is separated from most nearby cell groups by the FLM, and also because the uptake and transport of HRP by intact axons of passage seems to be negligible in the central nervous system (Mesulam, '82). It also seems reasonable to assume that among the FLM axons close to the oculomotor nucleus (and hence most likely to take up enzyme), those terminating in this nucleus greatly outnumber those terminating in more rostral nuclei. In the work described here, the cell groups constituting the greatest hazard of interfering in the results are the Edinger-Westphal nucleus (the parasympathetic component of the oculomotor complex) and the trochlear nucleus, both of which were partially or totally affected by the brown halo surrounding the injection site, even when the central dense core of the injection was confined entirely within the somatic component of the oculomotor nucleus. However, the area covered by the brown halo is generally considered to lie outside the region of effective HRP uptake (Mesulam,'82). In chicks in which the central dense core affected the Edinger-Westphal nucleus, the present results agreed with reports of projection to this nucleus from the contralateral pretectal area in the pigeon (Reiner et al.,'83;Gamlin et al.,'84); and the appearance of only a very few labelled neurons in the pretectal areas of the three chicks in which the central dense core was smallest and most accurately centered in the OCM suggests that the present results have not been seriously distorted by the trochlear nucleus.

[4] 668w In the monkey (Buttner-Ennever and Buttner, '78; Steiger and Buttner-Ennever, '79), injection of HRP into the oculomotor nucleus has been reported to label neurons in the contralateral abducens nucleus, the vestibular complex, the rostral interstitial nucleus of the FLM (mainly ipsilaterally), the interstitial nucleus of Cajal (mainly contralaterally), and the ipsilateral perihypoglossal complex, as well as smaller numbers ventral to the oculomotor nucleus, ventrolateral to the fasciculus longitudinalis medialis, in the contralateral nucleus pretectalis olivaris, and in the medullary reticular formation (the nucleus pontis centralis caudalis, nucleus paragigantocellularis dorsalis, and nucleus gigantocellularis). In the cat (Graybiel and Hartwieg, '74), injection of HRP into the oculomotor nucleus revealed projections from the interstitial nucleus of Cajal, the abducens nucleus, the nucleus prepositus hypoglossi, and the vestibular complex, apart from labelling a few additional neurons in the nucleus of the posterior commissure and a t the lateral border of the pontine tegmentum near the rostral pole of the trigeminal complex. The present findings in the chick exhibit both great similarities and certain differences with respect to those reported for mammals. Particularly noteworthy in this regard is the group of cells labelled in the ipsilateral rostral mesencephalon a t the level of the medial region of the nucleus campi Foreli. This group may be homologous to the mammalian rostral interstitial nucleus of the FLM and thus may be involved in the generation of vertical eye movements (Buttner e t al., '77; Biittner-Ennever and Buttner, '78; Steiger and Buttner-Ennever, '79; Nakao and Shiraishi, '83, '85), which in birds also have an important effect on variation in the extent of binocular overlap (Martin,'86a). Brecha et al. ('80) found that in the pigeon the nucleus of the basal optic root complex (nBOR, the main nucleus of the that the great mobility of the head had taken over the role of eye movements, which were largely ignored. In recent years, studies of various bird species (Bloch et al., '84; Martinoya et al., '84; Wallman and Pettigrew, '85; Martin, '86a,b; McVean and Stelling, '86) have nevertheless brought out the importance of the eye movements and their relation to the binocular and monocular fields. Several researchers (Bloch et al., '84; Wallman and Pettigrew,'85;McVean and Stelling,'86) have reported conjugate eye movements in response to objects appearing in the frontal visual field and unconjugate eye movements in response to objects appearing in the lateral field. It has been shown (Martinoya et al., '84) that feeding pigeons employ coordinated vergence of both eyes in response to stimuli appearing in the frontal binocular field, even when one eye is totally occluded, which implies central neurological control of binocular fixation. In their comparative study of saccadic eye movements in the little eagle Haliaetus mophonoides and the tawny frogmouth Podargus stigroides (the former species being specialized for frontal binocular vision and the latter for panoramic vision), Wallman and Pettigrew ('85) found that in both species saccades could be either conjugate or very different in amplitude or opposite in direction, but all measured saccades occurred synchronously in the two eyes. The difference between the two species was that the eagle's eyes moved in the same direction in 73 ' 3% of all saccades, and the percentage of approximately conjugate movements was greater than in the frogmouth, whose eyes moved in opposite directions in 63 % of saccades. Although there has been little research on eye movements of the chick, it is known (Wallman and Pettigrew,'85) that the frequency of saccadic movements in this species is similar to that found in the eagle and that all saccades start within 8 msec of a saccade in the opposite eye (90% within 4 msec). Finally, the existence of optokinetic nystagmus in birds has been long known (Ter Braak, '36), and it has recently been shown (Gioanni et al., '81) that monocular stimulation of pigeons produces a conjugate optokinetic nystagmus of both eyes (though the amplitude and velocity of the movement are greater in the stimulated eye). Similar behaviour has been reported for the chick (Wallman and Pettigrew, '85).

[5] 215w Although the studies discussed above have demonstrated the considerable importance of eye movement in birds and the existence of significant binocular yoking, little research has hitherto been carried out on the neuron structures responsible for organizing these movements. The present study contributes to filling this gap. avian accessory optic system) has connections with several centres affecting eye movements, including projections to the oculomotor nucleus that would provide a basis for very rapid visually triggered input to the oculomotor muscles. These authors suggested that the nBOR may have a role in retinal stabilization and/or in rapid orientation to visual stimuli. Wallman et al. ('81) have shown that the nBOR of the chick responds to retinal slip and that these signals have a role in optokinetic nystagmus. Projection from the accessory optic nuclei to the oculomotor complex has also been observed in the catfish (Finger and Karten, '78). In mammals, no labelled neurons have been reported in the medial terminal nucleus (which corresponds to the nBOR of birds) after injection of HRP into the oculomotor nucleus (Graybiel and Hartwieg, '74); Steiger and Buttner-Ennever, '79), although Giolli et al. ('88) have recently described a projection from the medial terminal nucleus to the dendritic field of the oculomotor neurons, which might have an equivalent role, in rats and rabbits.

[6] 133w It is known that in mammals, e.g., the cat and the monkey, the interstitial nucleus of Cajal projects to both the oculomotor nucleus (Graybiel and Hartwieg, '74; Steiger and Buttner-Ennever, '79) and the spinal cord (Castiglioni et al.,'78;Zuk et al.,'82), lending weight to the suggestion (Szentlgothai,'43;Hyde and Toczek,'62;Carpenter et al.,'70) that this nucleus is involved in the coordination of head and eye movements. A similar role may be hypothesized for birds, given their extensive use of rapid head movements for exploring their visual environment (Dunlap and Mowrer, '30; Frost, '78), the present observation of mainly contralat-era1 projection from the interstitial nucleus of Cajal to the oculomotor nucleus in the chick, and the reported indications of projection from the interstitial nucleus of Cajal to the ipsilateral spinal cord in the pigeon (Brecha et al.,'80).

[7] 270w An interesting feature of the present findings is the significant number of cells labelled in the medial mesencephalic reticular formation close to the interstitial and oculomotor nuclei and the oculomotor roots. No labelled neurons were reported in this area after injection of HRP into the oculomotor nucleus of the cat (Graybiel and Hartwieg, '74), although in the monkey (Steiger and Buttner-Ennever, '79) labelled neurons were found scattered singly or in small groups just ventral to the oculomotor nucleus or ventrolateral to the fasciculus longitudinalis medialis. Nevertheless, subsequent electrophysiological studies have observed convergence and divergence neurons intermingled in the mesencephalic reticular formation of the monkey just outside the oculomotor nucleus (Mays, '84); in the cat (Nakao et al.,'861, neurons that projected directly to the medial rectus subdivision of the oculomotor complex, and which were presumed to be involved in vergence eye movements, were identified bilaterally in the ventralmost part of the periaqueductal grey matter and in the medial mesencephalic reticular formation. These reticular neurons were located at the level of the interstitial nucleus of Cajal and the oculomotor nucleus, most lying ventral to the interstitial nucleus of Cajal, lateral to the Edinger-Westphal nucleus, or on the border of or outside the fasciculus longitudinalis medialis. In view of the proven importance of vergence eye movements in various bird species (Martinoya et al.,'84; Wallman and Pettigrew, '85), it seems reasonable to suppose that at least some of the neurons labelled in the present study in positions similar to those described above may likewise be involved in such vergence eye movements. It should nevertheless be borne in mind that very little is yet known

[1] 5w Eight-week-old Harco-red chicks (Gallus domesticus)

[2] 105w weighing 500-750 g were anesthetized by intramuscular injection of 30-35 mg/kg of Nembutal and supplementary doses of ethyl ether. In preliminary experiments, chicks were placed on a David Kopf stereotaxic instrument, and dye was injected (using a 1-p1 Hamilton syringe) to determine both coordinates of the oculomotor nucleus and an angle of entry such that the needle traversed the nucleus rostrocaudally. Subsequently, the oculomotor nucleus of 14 chicks was injected stereotaxically with 0.015-0.1 g1 of 30% HRP (Serva) in 2% dimethylsulfoxide over a period of 45-60 minutes by means of a l -~l Hamilton syringe or a glass micropipette connected to a WPI nanolitre pump.

[3] 168w After a survival time of 30-40 hours, the chicks were anesthetized again, heparinized, and perfused over a period of 30 minutes, first with isotonic saline and then with 800-1,000 ml of 2.5% glutaraldehyde in 0.1 M phosphate buffer at pH 7.2. Because washing and fixing the brain is much more difficult for long-necked animals such as the chick than it is for common laboratory mammals, perfusion was performed by transcardiac bicarotid catheterization. The brain was removed from the skull, the brain stem blocked transversely, and the blocks postfixed for 4-6 hours in the same fixative as before and then washed for 6-12 hours in 0.1 M phosphate buffer at pH 7.2. Transverse serial sections 75 pm thick were cut with a vibrotome, collected in phosphate buffer, and developed with tetramethylbenzidine (TMB) (Gomez- Segade et al.,'80;Mesulam,'82) or o-tolidine (dimethylbenzidine) (Segade, '87). Sections treated with TMB were left out to dry, cleared in xylene, and mounted without counterstaining, whereas those treated with o-tolidine were usually lightly counterstained with pyronin Y.

[4] 100w In sections treated with o-tolidine and pyronin Y, those cells with clearly visible nucleoli were measured with a Kontron Videoplan semiautomatic image analyzer. The parameters determined included the maximum diameter of the perikaryon (Dmaxj, the area of the perikaryon section (Ap), the diameter of the circle of area Ap (Dc), and the form factor of the perimeter (ff = 4 7r Ap/[perimeter]'), which approaches unity as the shape of the perikaryon section becomes rounder. In counting neurons, overcounts due to perikarya appearing in more than one section were corrected for by multiplying the number counted by Konigsmark's correction factor (

[5] 38w is the thickness of the sections (75 pm), r the mean value of the radii of circles with the same area as the perikarya counted, and k the diameter below which the perikarya are not measured (Konigsmark, '70).

[1] 101w Eye movements and their neurology have been extensively studied in mammals with frontal eyes and binocular vision, but relatively little is known about nonmammalian vertebrates, most of which have lateral eyes. In particular, the fact that most birds have lateral eyes and highly mobile heads that allow them to scan their surroundings by means of rapid head movements (Dunlap and Mowrer,'30;Frost,'78) led to their eye movements being ignored. In recent years, however, several researchers (Gioanni et al.,'81;Bloch et al.,'84;Martinoya et al.,'84;Wallman and Pettigrew,'85;Martin,'86a,b) have established the importance of such movements and the existence of binocular yoking in various species of bird.

[2] 208w Little is known of the neurological mechanisms controlling avian eye movements, despite the relevance of such knowledge to an understanding of the evolution of binocularity in vertebrates. In particular, little research has been done with tracer techniques. Retrograde transport of horseradish peroxidase (HRP) has been used to study the arrangement of motoneurons in the oculomotor nucleus of chick embryos (Heaton and Wayne,'83), the location of the motoneurons innervating the trochlear muscle of duck embryos (Sohal and Holt,'78), and the abducens, pyramidalis, and quadratus muscles of the chick '87). The abducens nucleus of the chick has also been found '87) to contain a large group of internuclear neurons projecting to the contralateral oculomotor nucleus that are morphologically similar to those of mammals; also, bilateral projections from the nucleus of the basal optic root complex to the oculomotor nucleus, interstitial nucleus of Cajal, and vestibulocerebellum have been described in the pigeon (Brecha et al., '80). However, control of the complex avian eye movements as recently reported must require the existence of further neuronal connections. The existence in birds of vestibular neurons projecting to the oculomotor nucleus was first reported at the beginning of the century (Wallenberg, 1900; Ram6n y Cajal, '08) and was later confirmed by Wold ('78) with HRP.

[3] 53w In this respect, there are, however, considerable discrepancies between Wold's findings and the present findings, possibly because Wold used a less sensitive histochemical technique and did not achieve strictly unilateral injections into the oculomotor nucleus. The present article describes a retrograde HRP study of afferent connections to the oculomotor nucleus of the chick.

[4] 119w about the organization of vergence eye movements in birds; in particular, it is not clear whether avian eyes execute slow, symmetric vergence movements similar to those observed in primates or only vergence movements controlled by the saccadic system (Wallman and Pettigrew,'85). The fact that greater numbers of neurons were labelled in this region when the injection site was large or the concentration of HRP in the ventral subnucleus greatest suggests that some of these neurons may have been labelled as the result of uptake of tracer by fibres crossing or bordering the ventral boundary of the oculomotor nucleus; an alternative explanation might be that some mediate movements involving the inferior oblique and/or contralateral superior rectus muscles (Heaton and Wayne,'83).

[5] 220w The large number of interneurons projecting to the oculomotor nucleus from the contralateral abducens nucleus, and their similarities and differences with respect to those of mammals, have been described in detail elsewhere (Labandeira-Garcia et al., '87). Because in cats and monkeys abducens internuclear neurons are known to be involved in conjugating saccades, the existence of such large numbers in the chick is quite striking. Possible hypotheses are that their action on the medial rectus motoneurons is weaker in birds than in mammals, or that in birds there are both excitatory and inhibitory interneurons, which might explain not only the weaker and more variable yoking of the eyes, but also the symmetric convergent and divergent eye movements observed in the frogmouth (Wallman and Pettigrew, '85). An additional possibility is that the neurons labelled in the reticular formation at the level of the caudal third of, or just caudal to, the abducens nucleus may inhibit the medial rectus motoneurons directly. In the cat, the inhibitory burst neurons found caudal to the abducens nuclei inhibit contralateral abducens motoneurons and internuclear neurons, and there is no direct inhibition of medial rectus motoneurons (Kaneko and Fuchs, '81); however, certain findings (Grantyn et al.,'80) suggest that a direct mechanism may have existed formerly, only to be lost upon the development of frontal binocular vision of mammals.

[6] 144w No group of labelled cells was found that appeared to be homologous to those found in monkeys and cats in the perihypoglossal complex (mainly in the nucleus prepositus hypoglossi). In the cat, the nucleus prepositus has both afferent and efferent connections, with numerous areas of the brain stem involved in eye movement; this has led to the suggestion (McCrea and Baker,'85) that this nucleus is involved in the construction and distribution of a central efference copy of motor commands for eye movements (par. ticularly horizontal movements). The feline nucleus prepositus also has a rather modest direct projection to the oculomotor nucleus. The function of this latter connection is not clear, but it has been suggested that it is a case of parallel processing of information in the oculomotor system (Baker and Berthoz, '75; McCrea and Baker, '85). It is not present in the chick.

[7] 420w The vestibular neurons labelled by injection of tracer into the oculomotor nuclei of cats and monkeys (Graybiel and Hartwieg,'74;Gacek,'77;'79;Carpenter and Carleton,'83;Carpenter and Cowie,'85) have largely belonged to the nucleus vestibularis superior, the nucleus vestibularis medialis, and cell group Y, with smaller numbers appearing in the rostra1 part of the nucleus vestibularis descendens and only a few scattered cells in the nucleus vestibularis lateralis. As to the nucleus vestibularis superior, the pattern observed in the chick is very similar to that found in these mammals, in which most labelled cells appeared at the centre of the ipsilateral nucleus (Gacek,'77;Carpenter and Cowie,'85) and smaller numbers at the periphery of the ipsilateral nucleus or in the contralateral nucleus. In the nucleus vestibularis medialis of cats and monkeys, the great majority of neurons projecting to the oculomotor nucleus projected to the contralateral nucleus (Graybiel and Hartwieg,'74;Gacek,'77;'79;Carpenter and Cowie,'85); in the chick this difference is even more marked. Neurons labelled in the nucleus vestibularis medialis of the monkey were described as arranged in an elongated formation stretching ventrolaterally from the floor of the ventricle into the inferior vestibular nucleus near its border with the lateral vestibular nucleus; cells observed in the territory of these three nuclei thus seemed to form a single entity (Steiger and Buttner-Ennever, '79). In the same way, labelled chick cells that form a band occupying the territories attributed to the ventrolateral region of the nucleus vestibularis medialis, the nucleus tangentialis, and the rostral part of the nucleus descendens do not appear to be objectively susceptible to sharp division among the three nuclei on the basis of location and perikaryon morphology (Wold,'76). For topographic purposes, however, most of the 625-700 labelled neurons may be considered to belong to the tangentialis and medialis nuclei, with overlap between the two in a large central area that also features a smaller number of neurons that may belong to the rostral tip of the nucleus vestibularis descendens. In sections through the rostral half of the abducens nucleus, most labelled cells lie toward the nucleus tangentialis territory, whereas more caudally the majority lie toward the territory of the nucleus vestibularis medialis. Wold ("78) divided the labelled neurons in this region between the nucleus tangentialis and the nucleus vestibularis medialis, thus modifying the territories that he had previously established by thionine staining and Weigert's method (Wold,'76), the nucleus tangentialis being considerably extended medially and the rostral tip of the nucleus descendens being shifted caudally. In the present study, no clear criteria could be determined for establishing such boundaries.

[8] 119w In monkeys (Steiger and Buttner-Ennever, '79; Carpenter and Cowie, '85) and in cats (Gacek, '77), a few bilaterally labelled cells have been reported in the rostral region of the nucleus descendens near its border with the medial nucleus and Deiters' nucleus. In the chick, as well as the above-mentioned nucleus descendens cells that form part of the contralateral ventrolateral band, a group of 120-130 ipsilateral neurons were observed just lateral to the lamino-olivary tract in an area of overlap among the nucleus tangentialis, the rostral tip of the nucleus descendens, and the nucleus medialis. The morphology and arrangement of this group of neurons suggested that they should be considered to belong to the rostral pole of the nucleus descendens.

[9] 440w It is unclear if the group of neurons labelled in the nucleus tangentialis has any counterpart in mammals. The most apparently likely homologue of the nucleus tangentialis, the nucleus interstitialis (Brodal and Pompeiano, '57); Sadjadpour and Brodal, '68), has not been found to project to the mammalian oculomotor nucleus to any significant extent, as the few cells labelled in this position in the cat were regarded as displaced nucleus medialis neurons (Gacek,'77). In view of the similarity, on the one hand, between the band of neurons labelled contralaterally in the chick in the nucleus vestibularis medialis, the rostral pole of the nucleus descendens, and the nucleus tangentialis, and, on the other hand, the formation described in the monkey (Steiger and Buttner-Ennever, '79) as a single entity affecting the nucleus vestibularis medialis and the rostral part of the nucleus descendens, it seems possible that the labelled tangentialis cells may correspond to the most ventrolateral area of the formation described in the monkey. Wold ("78) suggested that the nucleus tangentialis might be homologous to cell group Y, the third large group projecting to the oculomotor nucleus in mammals (the others being the nucleus superior and the nucleus medialis). Nevertheless, a more likely homologue for cell group Y would seem to be the group of cells found in the present study dorsal to the contralateral tractus spinocerebellaris dorsalis, because, like the mammalian neurons labelled in cell group Y (Gacek,'77;'79;Stanton,'80;Carpenter and Carleton,'83;Carpenter and Cowie,'85)' they possess large perikarya and are arranged as a band capping the tractus dorsally in a position dorsolateral to the nucleus vestibularis superior and the nucleus Deiters dorsalis and ventrolateral to the nucleus cerebellaris lateralis. However, there seemed to be no convincing reason to consider these cells as belonging to neighbouring nuclei such as the nucleus vestibularis superior or nucleus cerebellaris lateralis. Although the possibility that nucleus cerebellaris lateralis cells may project to the oculomotor nucleus has been suggested for many years (Carpenter and Strominger, '64; Ron and Robinson, '73; Chan-Palay, '77), it was not supported by HRP studies in cats and monkeys (Graybiel and Hartwieg,'74;'79;Carpenter and Cowie,'85). It thus appears (Carpenter and Cowie,'85;'87) that this nucleus projects to areas close to but outside the oculomotor nucleus, which is in keeping with the fact that in the present study of the chick, numerous cells were labelled in the nucleus cerebellaris lateralis when injections were performed just lateral to the fasciculus longitudinalis medialis and the roots of the oculomotor nerve. As in the monkey (Gonzalo-Ruiz et al., "9, most of these labelled nucleus cerebellaris lateralis cells appeared to be the result of uptake by the dorsomedial region of the red nucleus.

[10] 25w Finally, the absence of any significant projection from Deiters' nucleus to the oculomotor nucleus again parallels findings reported for mammals (Graybiel and Hartwieg,'74;'79;Carpenter and Cowie,'85).

[11] 37w It would now be desirable for electrophysiological studies or experimentally induced lesions to be carried out to determine the functions of the various neuron groups that project to the oculomotor nucleus of the chick as described above.