58 VISION 



cells and an internal layer of flat squamous cells. These cells were broadly 

 fused laterally, a situation similar to that of cyprinid fish. This differed 

 markedly from the situations in tortoise and man, where horizontal cells 

 make lateral contacts through cytoplasmic extensions. Since the horizontal 

 cells of the elasmobranchs are so broadly coupled (up to 50% of the lateral 

 area, according to Stell), Yamada and Ishikawa suggested that this cell group 

 is functionally a single unit, i.e., an electrical syncitium. Finally, they 

 described conical processes from the external row and long narrow processes 

 from the internal row of horizontal cells, which apparently contact receptor 

 pedicles. 



In a preliminary study, Gallego (1972) observed horizontal cells of the 

 nurse shark Ginglymostoma under the light microscope by impregnating 

 whole retinas with reduced silver. He had previously postulated (Gallego 

 1971) that axonless horizontal cells mediate interaction between rods and 

 bipolar cells, while horizontal cells with axons mediate interactions between 

 cones and bipolars. Because sharks were said to possess pure-rod retinas, 

 Gallego felt that the studies of these animals might be crucial to his theory. 

 The layer of horizontal cells in the nurse shark retina stained by reduced 

 silver were indeed axonless, a finding that agreed with Stell's and Kaneko's 

 observations on other elasmobranch horizontal cells. However, Gallego was 

 apparently unaware that Hamasaki and Gruber (1965) and Wang (1968) had 

 reported that cone photoreceptors make up about 12% of the visual cell 

 population in Ginglymostoma. Thus his findings on Ginglymostoma cannot 

 confirm his suggestion that rods are related to axonless horizontal cells. 



Stell and Witkovsky (19736) studied the structure and synaptic con- 

 nections of Golgi-impregnated horizontal cells of Mustelus under the light 

 microscope. They recognized three layers of horizontal cells, all of whose 

 processes reach the external plexiform layer (Figure 12). Cells of the distal 

 layer (HI) are massive, perhaps 200 X 125 X 25 urn, and each probably 

 sends processes into every rod spherule in its field. The flattened horizontal 

 cells of the intermediate layer (H2) form clusters of terminals, and each rod 

 within the field of a cluster may receive a contact. Thus each rod apparently 

 receives contacts from HI and H2 cells. Cells of the third, most proximal 

 layer (H3), are flattened and stellate in form. They send relatively thick 

 processes vertically and contact cones exclusively. This situation is reversed 

 in teleosts, where HI cells contact cones and H2 and H3 cells contact rods. 



Stell and Witkovsky never observed H3 contacts within rod spherules or 

 HI and H2 processes associated with cone pedicles. This does not rule out 

 the possibility of rod-cone cross-connections through horizontal cells via 

 photoreceptor telodendria. They noted that elasmobranch horizontal cells 

 with axons have never been observed. Anctil and Ali (1974) confirmed the 

 presence of three rows of horizontal cells in the retina of the hammerhead 

 Sphyrna lewini, but did not determine horizontal cell-photoreceptor rela- 

 tionships. 



Yamada and Ishikawa (1965) have reported specialized contacts between 

 horizontal cells similar to the so-called gap junction first described by Revel 

 and Karnovsky (1967). Gap junctions between horizontal cells of other elas- 



