576 Buller.—Contributions to our Knowledge of the 
Dicksonia antarctica , fifty to fifty-five minutes ; Blechnum 
occidental , thirty-five to forty minutes ; species of Alsophila , 
Polypodium , Asplenium , Neprolepis and Osmunda , twenty- 
five to forty minutes. My own observations on Gymnogramme 
Martensii differ considerably from those made by Stras- 
burger and Voegler for other spermatozoa. The swarm- 
period for this species is about 120 minutes. 
In determining the length of the swarm-period, I have 
used the hanging-drop method already described (p. 573). 
A single prothallium with few antheridia was placed in a 
drop. It was thereby possible to watch the liberation of the 
spermatozoa. Several antheridia usually burst almost simul¬ 
taneously. When it was thought that no more antheridia 
would burst the prothallium was allowed to remain in the 
drop. If the contrary was feared the prothallium was re¬ 
moved. 
The time intervening between placing the prothallia in 
water and the bursting of the antheridia varies according to 
the ripeness of the latter. In a considerable number of ex¬ 
periments the time was ten to twenty minutes. After an 
antheridium has burst several minutes, often about eight, are 
required before the spermatozoa have all become set free 
from the mother-cells. In a few cases individual spermatozoa 
after liberation from an antheridium did not swarm at all. 
Sometimes a few spermatozoa were left in an antheridium 
after the others had escaped and there continued to move for 
a considerable time. 
It was found that in tap-water the majority of spermatozoa 
move well for about two hours, after which they come to rest 
one by one, the longest time of movement being about three 
hours. This result is somewhat remarkable in view of the 
already quoted, much briefer swarm-periods given by Stras- 
burger and Voegler for other spermatozoa. In particular, 
Voegler’s estimation of the swarm-period for Gymnogramme 
Laucheana is only twenty-five minutes. According to my 
estimation the swarm-period for the nearly related G. Martensii 
must be considered at least five times as long. 
