696 



HANDB(50K OF PHYSIOLOGY ^ NEUROPHYSIOLOGY I 



can be divided into two types, large and small, with 

 two extremes, giant and midget. It is doubtful 

 whether the midget system occurs in the common 

 laboratory animals. The dendritic expansions of the 

 giant ganglion cells may be "from 250 to 350 n across 

 and probably more" (Polyak). 



The system of amacrine cells (i.e. cells without 

 axons) seems highly organized. This, according to 

 Ramon y Cajal, is particularK true for the stratified 

 ones (Polyak's knotty amacrines) which form five to 

 seven layers in the inner plexiform network. Their 

 dendritic arborizations are similarly stratified at cor- 

 responding levels. There are also giant amacrines 

 (Polyak's tasseled amacrines) some of which spread 

 a 'daddy longlegs' mop extending over i mm above 

 the plane of the ganglion cells. Polyak has also de- 

 tected axons running from amacrine cells towards the 

 pedicles of the receptors and regards them as bipolar 

 cells conducting backwards. This raises the question 

 of whether amacrines appear to lack axons merely 

 because of difficulties in staining. 



In this microcosm of a nervous center that we call 

 a retina, the inner plexiform layer, as we have seen, 

 is the meeting ground of three major systems and thus 

 a critical region. It is difficult to imagine this layer 

 to be wholly self-controlled lay the chance play of 

 light and shadow. And, as a matter of fact, this is the 

 very region to which the centrifugal fibers of Ramon 

 y Cajal (123) and Dogiel (50) were found to project. 

 Ramon y Cajal studied them in the retina of the dog 

 (fig. 3), while Dogiel worked on birds. They seem 

 to be difficult to stain and their origin is unknown, 

 yet Ramon \ Cajal did not hesitate to postulate a 

 central origin rather than to describe them as re- 

 current collaterals. Some centrifugal fibers are held 

 to go as far as to the outer plexiform layer. 



The briefest path in the retina clearly is disynaptic : 

 receptor-bipolar-ganglion. A more fundamental issue 

 seems to be the question of whether bipolar cells 

 make axosomatic or axodendritic connections with 

 the ganglion cells. Conduction is slow in dendrites 

 (104) so that axo.somatic latencies are likely to be 

 shorter. In the probable absence of midget cells in 

 the common laboratory animals, the size of the 

 ganglion cell is likely to be an important property 

 because the larger the cell, the greater the probability 

 of axodendritic activation in the inner plexiform 

 layer. Actually the ganglion spikes in the cat's retina 

 fall into two main categories, large and small, the 

 small ones as a rule having brief latent periods. The 



larger the spike caused by illumination, the later it 

 tends to be discharged and the lower its absolute 

 threshold to light. This is the author's general im- 

 pression, not a result of systematic analysis. 



If, in the cat's eye, one proceeds to send an anti- 

 dromic (backward) shock into the optic nerve and 

 places a microelectrode on the blind spot (74), the 

 volley recorded consists of an early large and a later 

 small group of spikes, similar to those recorded ortho- 

 dromically at the central end of the optic nerve (22, 

 23, 107). The maximal conduction velocities of its 

 fillers are 70 and 23 m per sec, respectively (22). Two 

 main fiber sizes (as judged from the conduction veloci- 

 ties) suggest two main groups of sizes of ganglion cells. 

 This is further evidence for subdi\'iding the spikes into 

 two main categories. 



At the blind spot the optic nerve loses its myelin 

 sheath and so conduction suddenly slows down as the 

 antidromic impulse enters the fibers running across 

 the retinal surface (74). Precise measurements by 

 Dodt (42) gave mean values of 2.9 and 1.7 m per 

 sec. for large and small spikes, respectively. The large 

 spikes (see below) are the ones most easily influenced 

 by centrifugal tetani (74) as also seems probable con- 

 sidering their wide dendritic expansions within the 

 inner plexiform layer. 



Another interesting point is that, on account of the 

 slow conduction across the retinal surface, the im- 

 pulses from the peripheral portions of the retina may 

 i^e delayed by 4 to 6 msec, as compared with those 

 arising in the region around the blind spot. This is of 

 technical interest because it means that, unless special 

 precautions are taken in studying retinal brain pro- 

 jections by evoked potentials, these are likely to be 

 mainly determined i)y the fibers around the blind 

 spot. Phvsiologically the delayed conduction means 

 that, with a moving retina, space coordinates stand 

 a good chance of being transformed into time co- 

 ordinates. The eye always makes small oscillations 

 in fixation (39, 125). 



ELECTRORETINOGR.KM (eRG) 



The electroretinogram is a polyphasic mass re- 

 sponse (fig. 4) with specific cornea-positive deflections 

 at the onset and cessation of illumination. Standard 

 leads in electroretinography are between the cornea 

 and an 'indifferent' point on the body or behind the 

 bulb (in the case of eyes excised from cold-blooded 



