KOPOID AND SWEZY: UNARMORED DINOFLAGELLATA 163 



its onion-skiu pink, is totally unlike the colors foiiud elsewhere in the Gymno- 

 diuiidae (pi. 3, fig. 51). Gumnodiniiun coerulcam is one of the few species 

 showing a clear bine color. Other species sometimes show a faint bluish tinge, 

 as occasionally in G. gracile and G. canus. It is noteworthy here that those 

 species having colored pigment aggregated into clumps have the shades of red, 

 purple, and violet found in Gi/rodiitiuin and in some of the most highly special- 

 ized forms in CocModinium, Pouchetia, and ErijtJtropsis. These colors are 

 found in the most generalized group of the genus Gymnodinkim. On the other 

 hand the vic\\ orange yellow found in the most highly specialized members of 

 that genus is entirely lacking in ErytJiropsis and Pouchetia and nearly so in 

 CocModinium. The line of development, in so far as determined hj the color, 

 leads onward from the more generalized species of the genus and not from its 

 most highly specialized ones. 



The species of this genus are in part holophytic and in part holozoie. The 

 l^redominant ij^e of nutrition in the more generalized subgenus Gymnodinium 

 is holophytic. the majority of species in that group i:)ossessing chromatophores, 

 yellow, yellow ochre, and green in color. In some of these forms, however, the 

 possibility of holozoie nutrition, even when chromatophores are present, is not 

 altogether excluded, as shown by the presence of foreign bodies in the cytoplasm, 

 as in G. agile (fig. Y, 9), G. herlxiceum (fig. Y, 17), G. ravenescens (fig. X, 21), 

 and G. favum (fig. X, 7). This condition is similar to that in Amphidinium. 

 steini (Stein, 1883, pi. 17, figs. 14-16) and A. scissum (pi. 2, fig. 22). In the 

 species G. fidgens (Lebour's (1917&) G. pscudonoclilHca, our fig. X, 30) several 

 large bodies are present, but no intimation is given as to their nature. The 

 cA'toplasm of G. hcrhaceiim particularly is filled with the accumulated products 

 of metabolism in the form of oil droplets and green, blue, and grey vacuoles. 



In the next two siibgenera, Lincadinium and Pachijdinium, nutrition is 

 probably largely, if not entirely, holozoie. In the first named group indisputable 

 evidence of this is foimd in one species only, G. lieterostriatum. The individuals 

 of this species observed under the microscope proved to be insatiable cannibals. 

 The actual process of ingestion was not ol)served, but very few specimens were 

 noted which did not contain large food masses. In many cases these were 

 recognizable as species of Gymnodinioidae (pi. 5, fig. 56). Their ejection from 

 a ]:)osterior vent in the l)ody was often seen. In many of the other species the 

 cytoplasm usually contains oil droplets, vacuoles and refractive bodies, the 

 accumulations of the products of meta])olism suggestive of holozoie nutrition. 

 These are especially striking in the subgenus Pachydinium. The body here is 

 frequently filled with large food masses, as in G. puniceuni and G. sphaericum. 

 The sulcus in the midregion of the body probably functions as the mouth, and 

 the posterior part of the body, particularly the sidcal notch, as the vent for 

 ejecting foreign particles and fecal masses. 



Cyst fonnation is common in Gi/iuiiodiiiiinii. and the cyst generally takes 

 the form of a verv thin, hvaline mcmljraiu'. A second or even third membrane 



