DKNDKOSUWA HADIAKS, EKKENBERU. 
165 
seen any evidence of division of the niicronucleus during' tlie 
free-swimming gemmula stage nor in younger dxed stages. 
It seems probable, tlierefore that the number of micronuclei 
takiug pare in the formation of the bud may vary from one 
to seven. In JJendrocom etes we also found that the 
number is not constant, but varies from two to five. 
In JJendrosoma then, as m Dendrocometes, the mega- 
nucleus of the gemmula is formed by amitotic division of tlie 
megauucleus of the parent, and the micronuclei of the gemmula 
are formed by mitotic division of the micronuclei of the 
parent. 
The Fkee-Swiiiming Gemmei.a. 
We have called attention to the fact that there is consider- 
able variability m the size of the gemmulte of Deudrosoma. 
The gemmula shown m fig. 1 was plano-convex in form, with 
a broad girdle of several rows of cilia extending from the 
middle almost to the edge of the plane surface, it was 6Up 
ill diameter and 4U p in height, it was difficult to count the 
number of contractile vacuoles or to be certain they were 
constant in number, but there were certainly more than 
tliree, and probably from eight to ten. Their rhythmic con- 
tractions were not synchronous, and frequently, but not con- 
stantly, two or three vacuoles in close contact made their 
appearance when previously only one was seen. 
The megauucleus in this form could be clearly seen in the 
centre of the protoplasm when the gemmula Avas viewed from 
above or below, it was sometimes spherical, but varied in 
sliape, and frequently showed one or two lobate processes 
like tlie pseudopodia of an amoeba. 
This particular gemmula was kept under observation lor 
two days, and gave rise to a young suctoriau, which was 
clearly a Deudrosoma radians. 
The Iree-swimming gemmula; of the second kind (p. iOl), 
found ill the material from the iiridgew'uter canal, are pro- 
bably the gemmula of the Leruteophrya forms. They have 
