hours later) varied from 33,000 to 600,000 bacteria per ml. A 

 covmt of 24. million bacteria per ml was obtained after 31'2 hours 

 from the initially bacteria-free G.brevis cultiire in which one 

 fish survived the test period. The first bacterial counts for 

 two containers of initially sterile culture medium were 1,000 

 (container 5) and 25,000 (container 11) per ml. When these con- 

 tainers were sampled again near the end of the 31^-ho-\jLr test period 

 the bacteria had increased to 21 (container 5) and 7 (container 11) 

 million per ml. 



The results of experiment 8 show clearly that bacteria-free G. 

 brevis cultures axe toxic to fish. Nevertheless, we desired to con- 

 firm this toxicity using test materials with greatly reduced initial 

 and terminal bacterial counts. 



Experiment 9. Effects of Bacteria-free and Unialgal Gvmnodinium brevis 

 Cultures, and Effects of Filtration on Toxicity 



The second study with bacteria-free G. brevis differed somewhat 

 from the first one (Experiment 8). In experiment 9 the initial 

 bacterial contamination of test materials by containers and aeration 

 equipment was reduced; the effects of two methods of filtration on 

 the toxicity of bacteria-free cultures were studied; and the sensitivity 

 of two size groups of mullet ( Mugil cephalus ) to G. brevis cultures 

 was compared. The experimental setup was the same as for experiment 8 

 (Fig. 1) except that more containers and a larger air pump were used. 



We employed three precautions to reduce the initial bacterial 

 contamination of the test materials. One precaution was heating the 

 experimental containers (2-liter beakers) in a hot air oven at 

 150-160 C. for 2 hours and allowing them to cool overnight in the 

 oven. The containers were removed from the oven shortly before the 

 test materials were added. Secondly, the aeration apparatus was 

 autoclaved for 15 minutes at 15 pounds press-'jre. Before sterilization 

 this apparatus was assembled and packaged so that a glass air-delivery 

 tube could be inserted into each of the 18 containers without handling 

 the portion which contacted the test materials. All air pumped into 

 the containers passed through a non-absorbent cotton filter installed 

 in the main air line from the pump. The third safeguard against 

 excessive bacterial contamination was placing the test fish in 

 autoclaved 85^ aged sea water for several minutes before transferring 

 them to the experimental containers. This concentration of sea water 

 is about the same as that in the culture media to which the fish were 

 exposed. 



Thi-ee different month-old bacteria-free G, brevis cultures, 

 two batches of sterile culture medium of the same age, a 10-week-old 

 iini-algal G. brevis culture, and autoclaved S5% aged sea water consti- 

 tuted the test materials. One liter of test material was placed in 



25 



