188 DISCOVERY REPORTS 



The large size of the sections and the length of the ribbons made it clear at once that 

 such a method would add little to the general account of the anatomy given by Liiders 

 (1909). On receiving a second supply of many fine specimens collected by the R.R.S. 

 'Discovery II', which were fixed in formalin and preserved in alcohol, I decided to 

 try a celloidin method. The impregnation was very lengthy, but it was found in the 

 first few specimens that the thin dorsal dome of the trunk collapsed even though this 

 took several months. This collapse, however, did not interfere with the anterior region 

 and these series were of considerable use. Thus the series of photographs on Plate XL 

 showing the complete nerve ring are from one of these early series. To avoid the distor- 

 tion I cut away a small window in the shell dorso-laterally and then, through the hole 

 so made, another in the dorsal body wall. This allowed the celloidin to enter the body 

 cavity without any collapse of its walls and resulted in two series of sections in which 

 there appears to be no distortion at all (see Plate XLII). 



The celloidin blocks were cleared and cut under cedar-wood oil. This allowed of accurate 

 orientation. The sections were cut very thick, the series ranging from 70 to 120/i. 



The anatomy was worked out by the projection-reconstruction method, and I mention 

 this method particularly as I feel that it is too little used. The image of a section is 

 projected through the microscope on to a sheet of paper. In a thick section it is possible 

 to focus on to the upper surface and then on to the lower. For the first section both these 

 outlines are drawn. The next section is then projected and of course it is found that the 

 upper surface of this section fits accurately on to the outlines of the lower surface of the 

 first section. When the image of this second section has been thus accurately fitted into 

 the drawing of the first section, a drawing is now made of the lower surface of this 

 second section. The succeeding sections are treated similarly, and in this way a series 

 of contours is built up from which the shape of the organism sectioned can be accurately 

 deduced. 



It will be obvious that for this method there is an optimum thickness of section. If 

 the sections are too thick there will be too few contours so that an insufficient idea of 

 the solidity will be obtained, whereas if they are too thin inaccuracy is bound to result 

 in superimposing successive outlines, which will differ little from each other. I have 

 stated above the thicknesses which I used for this relatively large crustacean, but the 

 same method can be used for smaller forms, using much thinner sections and greater 

 magnification. 



I consider the method very accurate. All the figures in this paper, except where 

 otherwise stated, have been drawn by this method. I found that it was possible to 

 obtain much more accurate representations of the actual limbs from sections than from 

 camera lucida drawings of whole limbs. The view of any such reconstructed limb is 

 naturally determined by the plane of the section, but with three series of sections in 

 planes at right angles to each other three views of the same limb can be obtained which 

 together should represent its whole structure. 



An accurate superimposition of the image of one section on the drawings of previous 

 sections is essential for a successful reconstruction. This becomes immediately apparent 



