838 THE AMERICAN NATURALIST. [Vou. XXXII. 
insects; or they may be terminal or lateral, as in either the 
simple or the compound eyes of arthropods. But in all cases, 
except the small horizontal rods of Acilius, the long axes of the 
rods are parallel to the rays of light falling on them, as shown 
in Fig. 1, where the light in each instance is supposed to come 
directly from above. 
(3) The Shape of the Rods in Cross-Section and the Arrange- 
ment of the Retintdial Fibrils.— The shape of the rods in cross- 
section is a matter of great importance, because it usually indi- 
cates how the retinidial fibrils are arranged. The cross fibrils 
lie in superimposed planes, generally placed in one of the fol- 
lowing positions: (2) When the rods are cylindrical, the cross 
fibrils radiate from the centre of each rod, like the bristles of 
a test-tube cleaner (Fig. 8, A, Pecten). (4) When the rods 
are quadrilateral in section, as in Acilius and Lycosa, all the 
cross fibrils in the same transverse plane are nearly parallel 
(Fig. 8, B, C, D). (c) When the rods are bound together in 
groups of from three to eight, as in the compound eyes of many 
arthropods (Tabanus, Æ, Bdellostoma, 7, Dytiscus, G, Fig. 8), 
the angular relations of the cross fibrils will vary with the num- 
ber of cells in the retinula. (æ) In some crustacea (Penzus) 
the rods are serrated and dovetailed into one another in such 
a way that certain cross fibrils in a given transverse plane are 
parallel to one another, but at right angles to those in the 
planes above and below them (Figs. 1‘ and Fig. 8, Æ). 
But whatever may be the position or shape of the rods, their 
retinidial fibrils always stand at right angles to the rays of 
light that fall on them. This is shown in Fig. 1, where all the 
retinal cells are seen in their natural positions. Over each 
figure is a cross-section of the rods, showing the direction of 
the retinidial fibrils. These figures, however, give only a very 
rough idea of the number and delicacy of the fibrils. The 
modifications of the shape and of the position of the rods that 
may occur without changing this arrangement of the fibrils are 
very curious and interesting, but they cannot be detailed here 
beyond what is given in the figures. i 
(4) Position of the Retinidial Fibrils in Different Ocelli of the 
Same Animal. — Where there are several pairs of ocelli in the 
