How to obtain it. 127 



pure to do without. But this is a very rare exception. At or 

 near the point where it enters the hatchery then, construct the 

 niters. For pure spring water half-a-dozen or more flannel screens 

 will usually be found ample, and~oiYeri" three or four wnTdo. It 

 is better to have too many than too few, for they play a very 

 important part in the success of the work. For fifteen years 

 I have worked with no other filter, and now that a much larger 

 volume of water is required in the hatchery the same method is 

 essentially successful, except that the water is first passed through 

 a couple of settling tanks, which are found very useful adjuncts. 



Fig. 6. 



A simple filter shown in Fig. 6 explains itself. It consists of 

 a wooden box, six wooden frames with coarse flannel stretched 

 on them, sliding into groves at a moderate angle, an inlet and an 

 outlet, and the whole charred inside. The size depends entirely 

 upon the amount of work to be done and the state of the water. 

 As an example, I may say that I have incubated successfully half 

 a million ova in the water discharged through a set of four 

 (occasionally increased to five) flannel screens, of about seventy 

 square inches each. A double set of these (for convenience 

 in cleaning), each in a separate box, is used, the whole water 

 sometimes passing through one box, but, as a rule, both boxes 

 working. 



The filter boxes I have at present in use are twenty-four 

 inches by twenty-four inches, and two of these boxes now working 

 will pass 200,000 gallons of water per day, or enough to incubate 

 four millions of ova. Behind each filter are two settling tanks 

 built of concrete. This applies to the main hatchery only, two 



