l8 9i-l North American Hyphomycetes. 203 



may be inferred in Gonatorrhodum from the large basidia or 

 " ramuli," as well as from the spore of the primary chain from 

 which a secondary chain occasionally springs. The present 

 genus is purely Mucedinous; while, as Saccardo remarks, 

 Gonatorrhodum seems rather to belong among the Dematieae. 



Desmidiospora nov. gen. — Spores of two kinds, on the 

 same mycelium of hyaline septate hyphae: microconidia, small, 

 hyaline, subfusiform, produced at the apex of subulate lateral 

 basidia: macroconidia very large, terminal, brown, flat, multi- 

 locular, several times successively, more or less irregularly, 

 dichotomously lobed. 



Desmidiospora myrmecophila n. sp. — Plate XX, figs. 1-9. 



Hyphae much branched and septate covering the host in a 

 white flocculent mass. Microconidia subfusiform, slightly 

 asymmetrical 12x2-2.5/*, produced sideways at the apex 

 of subulate basally inflated basidia. Macroconfdia terminal, 

 short stalked, three to five or even six times successively 

 dichotomously lobed, irregularly multilocular, very thick 

 walled, reddish brown or fawn colored, 8ox68/i, maximum 

 100x90^. 1 2-14^ in thickness. 



On a large ant. Connecticut. 



This remarkable plant has been met with only once growing 

 luxuriantly on a large black ant which was found fastened to 

 the under side of a rotting log. The hyphre emerged especially 

 from between the abdominal segments, enveloping the insect 

 more or less completely and extending a short distance over 

 the substratum. It is not impossible that this may be an im- 

 perfect form' of some Cordyceps, perhaps C. unilatcralis Tul., 

 which is the only species of the genus that the writer has 

 observed on ants in this vicinity. No analogous form, bear- 

 ing the same relation to Cordyceps that Mycogone or Sepe- 

 donium, for instance, are supposed to bear to other Hypo- 

 creaceous genera, has been observed, as far as known to the 

 writer. The microconidia in the present instance are, how- 

 ever, certainly very similar in appearance and mode of forma- 

 tion to conidia known to be associated with certain species of 

 Cordyceps: yet the connection cannot be assumed and the use 

 °f a new generic name seems unavoidable. It is not unlikely, 

 moreover, that the fungus under consideration may be myco- 

 philous and like certain species of Gymnoascus, Thiclavia, Me- 

 lanospora, etc., which so often interfere with artificial cultures, 

 b e parasitic on an immature Isaria or Cordyceps previously 

 developed within the insect. 



