84 EDWARD M. NELSON ON DIATOM STRUCTURE. 



becoming an integral portion of the central perforated membrane. 

 You will notice that the large peripheral secondaries are in all 

 stages; some are only just beginning to notch, others more 

 advanced are nearly cut in two. 



That something of this kind takes place, the figures in PI. 18, 

 Vol. 3, Her. 2, p. 201 (1888), in illustration of my paper on the 

 ^' Formation of Diatom Structure," go to prove ; further, Figs. 2, 

 4 and 9, PI. 8, and Fig. 2, PI. 20, Vol. 4, Ser. 2, p. 316 (1891), 

 illustrate the same thing, and are especially interesting because 

 they are the first stages in the formation of the delicate per- 

 forated cap of the Asteromphalus ; these may be found forming 

 a graduated series from the elementary triangle up to the 

 Asteromphalus pattern, as generally known, and as figured by 

 Mr. Karop in Fig. 1, PI. 17, Vol. 2, Ser. 2, p. 270 (1886). 



The evolution of the peripheral dots from the triangle is inter- 

 esting. I am now able to give a more complete account of this, 

 owing to the discovery of some intermediate forms since my 

 previous paper was written in 1891. 



In the first instance, for a termhius ad quern, we have merely 

 the plain polygonal (practically speaking hexagonal) structure with 

 the usual eye-spot layer attached. Next we find an equilateral 

 triangle at each intercostal, the apices of the triangle pointing 

 to the centre of the hexagons, and the sides of the triangle 

 cutting the sides of the hexagon at right angles (Fig. 2, PI. 8). 

 Next we find the triangle growing larger and the apices of it 

 becoming blunted. Next a large dot is formed between the 

 parallel sides of two adjacent triangles at a point about half 

 way between the intercostal points. By this means six large 

 perforations are formed in each hexagon, not at the angles of the 

 hexagon, but at the bisection of its sides (Fig. 9, PI. 8). About 

 this period the blunted apices of the triangle become notched, 

 the notches deepen, and eventually become perforations. These 

 perforations, which are at the intercostal angles, are not so large 

 as those at the bisection of the sides of the hexagon. At this 

 point, then, we have twelve perforations round each hexagon, six 

 large ones at the bisection of its sides, and six small ones at the 

 intercostal angles. At the next step the small intercostal per- 

 forations become elongated, then notched, and eventually divide 

 into two. The microscopical resolution of these is precisely 

 similar to the splitting of close double stars with a telescope 



