EDWARD M. NELSON ON DIATOM STRUCTURE. 83 



If further evidence be required, it may be remarked that six 

 spots with a central dot is the interference image or ghost of a 

 quincunx pattern. Now, the images in question cannot possibly 

 be interference images of the quincunx pattern of the primary 

 polygonal structure of the valve, because the position of the 

 intercostal ghosts would be in the intercostal areas ; again, as a 

 dernier ressort, it may be said that it is an interference ghost of 

 the thirteen (more or less) spots at the periphery of the cell cap : 

 to this it may be replied that any one making such a statement is 

 bound to prove that the interference ghost image of thirteen 

 (more or less) spots in a circle is the same as that of a quincunx 

 pattern ; w^hich of course could not be done. The crushing argu- 

 ment is, however, this : viz., that there are structures precisely 

 similar in every respect, both as as to size and arrangement of 

 spots, etc., which do not possess this tertiary structure. The 

 tertiaries cannot,- therefore, be interference effects arising from 

 the primary or secondary structures, otherwise they would be 

 seen in all cases where the conditions were similar. This brings 

 down the matter to two points : viz., that this tertiary structure 

 is what it appears, or is an interference ghost of some as yet 

 undiscovered quincunx membrane inside the polygonal cell. 

 Against this latter supposition we have the dissimilarity in the 

 patterns and number of tertiary spots. Fig. 3, PI. 8, is draw^n from 

 another diatom to show irregularity in pattern. There are, it 

 will be noticed, six, five, four, and three dots round a very faint 

 and ill-defined central spot. This central spot may or may not 

 be an out-of -focus image of the eye spot on the lower membrane ; 

 anyhow^, this is of no importance, for the argument rests on the 

 peripheral dots. No one will contend for a moment that such an 

 iiTegular pattern is merely an interference effect, and not the 

 image of an entity; its very irregularity negatives the idea. 

 Moreovei', it should be noted that this is a coarse structure. 



Now^, with regard to the meaning of this tertiary structure, 

 my theory is that it is a stage in the evolution of the central 

 perforated membrane ; the large spots at the periphery become 

 divided up and thrown into the central portion, while other large 

 ones are formed to be broken up in their turn until the valve has 

 reached its adult size. 



Fig. 7, PI. 8, shows this going on, for we have the large peripheral 

 secondaries breaking into two dots, and the tw^o dots thus formed 



