STUDIES ON AVIAN H^MOPEOTOZOA. 697 



On more tliau one occasion I have noticed the coinmenciui>- 

 formation of a clump in a cover-slip preparation of living- 

 parasites^ where every field contained numerous individuals. 

 Here and there are small numbers of parasites, which have 

 become entangled by their flagella, the distal portions of 

 which appear to be inextricably intertwined. 



Once started, the iiicrease in size of cluster may take 

 place in two ways : (1) by the addition of fresh individuals 

 from the surrounding medium, which are continually being- 

 attracted ; and (2) by the multiplication of forms already 

 present. The increase is undoubtedly due much more to the 

 former method than to the latter ; during the early stage, at 

 any rate, it is probably almost entirely due to the accession 

 of more individuals. In short, these clusters are formed 

 mainly by agglomeration. As a matter of fact, dividing forms 

 are comparatively rare in all the clusters I have examined 

 (cf. figs. E-G, PL 5). I once left a cover-slip prepai-ation 

 containing a great many free, active prirasites for two or three 

 hours; when I returned to it I found several large clumps 

 which had not been there before. It was impossible that these 

 rosettes could have arisen otherwise than by agglomeration ; 

 they all had their flagella centrally directed and resembled 

 the cluster of tig. g, except for the fact that some were even 

 larger. 



An early stage in the formation of a cluster is seen 

 in the micro-photograph reproduced in fig. E. The individuals 

 composing it differ appreciably in form and size; some of 

 them, at the periphery, had apparently only recently been 

 attracted, and were not yet firmly attached. Only two 

 individuals are undergoing division. The beginning of a 

 secondary agglomeration is instructive. Parasites continue 

 to be attracted to the clump, but owing to the number already 

 present the newcomers are unable to penetrate in between 

 them and become firmly attached. Hence they tend to form 

 a subsidiary cluster for themselves (figs. E and f). The large 

 agglomeration-cluster of fig. g is apparently made up of 

 individuals attached around three centres, two of which, the 



