FISHERIES OF ALASKA, 1908. 43 



great bulk of the run was, of course, estimated on the basis of this 

 quarter-hour tally, since the main part of the run came with a rush. 

 The daily tally for ten days of the season, nine of which were con- 

 secutive, was more than 100,000, and on July 14 more than 400,000 

 passed. This was the greatest number for any one day. During 

 both the earlier and later days of the season the daily run was meas- 

 ured by hundreds, or a very few thousands. During the height 

 of the run the tally coiild on clear nights be continued until midnight. 

 During these few days when the salmon accumulated in great num- 

 bers, faster than they could pass through, the body of fish imme- 

 diately below the web must have numbered several hundred thou- 

 sand. Large sections of them, alarmed at some cause or without 

 cause, would occasionally stampede and carry with them a con- 

 siderable wave lashed into foam, sometimes nearly across the channel, 

 accompanied by a characteristic roar caused by the churning of the 

 water. The fish did not run with entire uniformity tlu-oughout the 

 day. They usually dropped back as twilight or darkness approached, 

 and ceased to enter the gate and tunnel for a few hours. Sometimes 

 they would stop entering the trap without apparent reason and pass 

 by it back and forth along the web. In a heavy sea the tunnel 

 swayed somewhat and would not work evenly. The fish prefer a 

 rigid apparatus to pass through. 



Accuracy of the count. — The accurate estimation of a large run of 

 salmon such as enters the large rivers of western Alaska was con- 

 fessedly an experiment, and considerable doubt was felt as to whether 

 it would prove practicable. Not only w'as it necessary to make a 

 barricade which would certainly prevent the escape of salmon save 

 at the selected gates where the observations were to be made — in 

 itself no inconsiderable undertaking — but the salmon must be led 

 rapidly through these gates, and if not individually counted, at 

 least estimated by some method which commended itself as likely 

 to give a total so closely approximate to the real number as to be as 

 useful as an actual count, and preferably a method in w^hich the 

 result does not depend on the judgment of the observer. Elxperi- 

 enced fishermen can judge more or less truly with the eye the num- 

 ber of individuals in schools of fish, but there is error of the personal 

 equation in this judgment where there can be no subsequent count. 



Both the requirements referred to were fulfilled in the Wood River 

 investigations. Had the first heavy impulses of the run reached 

 the lake in the latter days of June, an important fraction might have 

 escaped unobserved past the rack, fcjr it had not yet been made a 

 perfect barrier, although even then it delivered most of the arriving 

 fish through the regular gate. A consideration of the dates of the 

 first heavy strike at the mouth of Nushagak Bay, of the first heavy 

 arrivals at Lake Aleknagik, and the known condition of the web, as 



