FISHERIES OF ALASKA, 1908. 41 



Since in several places the chain even at the base of the piles would not 

 remain on bottom, the chain was driven down by long timbers, 

 notched on the lower end and nailed to the pile above water to hold it 

 in place. By these means the salmon rack was made perfectly tight 

 and prevented entirely the passing of the salmon save through the 

 gates provided for the purpose, and by frequent observations A\dth the 

 water glass from the stern of a boat it was seen to remain in this 

 condition. 



The web across the channel was for the purpose of stopping the 

 ascent of the fish and at the same time to leatl them toward and into 

 the trap at the north shore. This trap was in water shallow enough 

 for all the fish in it to be seen, permitting, therefore, the estimating of 

 fish wdthin the trap or counting them as they passed out. The 

 salmon did not at first lead very readily to the trap, but played about 

 nearer the opposite shore. The entrance to the trap was at first a 

 mere gate 4 feet wide, but as the salmon used this gate to leave the 

 trap as well as to enter it, a tunnel several feet long was attached to it 

 reaching toward the middle of the trap. This effectually prevented 

 the return of any fish. The outlet gate was a rectangular opening 

 about 15i inches wide, and reaching (July 11) about 3 feet beneath 

 the surface. By the end of the season the fall of the lake level had 

 brought the water almost to the sill of the gate. This gate was pro- 

 vided with a door sliding in grooves, wliich could be raised or lowered 

 instantly, giving any desired aperture. During the height of the 

 season two simple tunnels were inserted directly in the rack web not 

 far from the trap. These were necessary to release the rapidly increas- 

 ing numbers of salmon below the web. The tunnels could be readily 

 closed by raising their mouths above water. 



The pile driver used in the work was kept at the lake until the end 

 of the season. It was moored close to the gate leading from the trap 

 so that observations on the salmon in the trap and on those passing 

 the gate could be made from the deck of the pile driver. The ark or 

 house scow was moored alongside the trap. This was the living 

 quarters for the attendants. It was close to the pile driver, with 

 which it was connected by foot planks. The tunnels were operated 

 from the deck of the ark. 



Method of counting and estimating. — The trap or corral was built 

 with the intention of using it as a basis of estimating numbers. It 

 was intended to let it fill with salmon, close the inlet, and tally out 

 the contents, repeating this one or more times until the observer 

 learned to estimate a trap full with approximate accuracy. Five 

 thousand salmon were found to pretty well fill the trap, but it was 

 not necessary to depend upon this method at all, and it would have 

 proved totally inadequate to dispose of the salmon during the mid- 

 dle of July. 



