278 BULLETIN: MUSEUM OF COMPARATIVE 790LOGY. 
pierces the basement membrane (mb. ba.) and extends to the first optic 
ganglion. Here it probably terminates in a fibrillation, as has already 
been shown to be the case in Astacus (Parker, ’95, p. 41). The acces- 
sory pigment cells (Fig. 1, el. sn.) occupy the space in the deeper part 
of the retina. The number of these cells is not constant, but, judging 
from their nuclei, it is not more than one or two for each ommatidium. 
Proximal processes extend from these cells through the apertures in the 
basement membrane to the distal surface of the first optic ganglion, and 
distal processes may extend forward: to the front faces of the distal 
retinular cells. Each ommatidium in Palsemonetes, then, is composed 
of the following cells: two corneal hypodermal cells, four cone cells, two 
distal retinular cells, eight proximal retinular cells (one of which is 
rudimentary), and a variable but small number of accessory pigment 
cells. Black pigment granules are contained in both the distal and the 
proximal retinular cells, and are limited to these cells ; the whitish pig- 
ment lies exclusively in the accessory pigment cells. The seven func- 
tional proximal retinular cells are the only elements of the ommatidium 
that are known to have nervous connections. These brief anatomical 
statements may suffice as an introduction to the consideration of the 
pigment changes in the retina. 
PHOTOMECHANICAL CHANGES IN NORMAL RETINA. 
The general method by which the normal photomechanical action of 
the retinal pigment cells in Palamonetes was determined consisted in 
the examination of eyes that had been kept in the light or in the dark 
known periods of time. For a dark chamber T used a box with a tight- 
fitting cover. From time to time during the course of the experiments 
this box was tested for its light-proof qualities by exposing in it a very 
sensitive bromide paper, such as is used by photographers. In all my 
experiments this showed complete absence of light. The top of the 
box was pierced by a hole, through which a piece of rubber tubing was 
introduced so that fluids could be poured into the box without exposing 
its contents to light. Two or three turns in this tube were found suffi- 
cient to prevent such light as entered the outer end of the tube from 
reaching the interior of the box. Living shrimps in a vessel of water 
were placed in the box, and the cover was carefully closed. After the 
expiration of the required interval, hot water was run in through the 
tube, and the animals were thus killed in the dark. Other killing re- 
agents, such as corrosive sublimato, picric acid, ete., were tried, but 
