82 VICTOR E. SHELFORD. 



rising-stem straight-way valve, with hose end nipple attached by means of a j-in. 

 brass ground joint union. 



The experimental boxes (Fig. lA) for the gradient experiments were designed 

 by the senior author and were 6.5 cm. wide by 30 cm. long, and 2.5 cm. deep. The 

 animals were confined in a portion of each box, 5 cm. wide, by a screen (8 meshes 

 to the centimeter). Air was supplied to the boxes by means of three fish-tail burner- 

 shaped introducers, opening into a narrow slit along the rear midway between top 

 and bottom. The area of the slit in each introducer was approximately the same 

 as the area of the 7 mm. brass tube which connected it with the pump. The air 

 after leaving the fishtail burner-shaped introducer passed through the screen that 

 confined the animals away from the slit, across the box, and out of the front through 

 a screen similar to the first. The air tended to spread out at about the angle of 

 the side of the introducer, and a small piece of metal was inserted between the rear 

 wall and the confining screen, to deflect this part of the flow directly across the box. 

 Each experimental box rested upon a movable board to each end of which a ring 

 stand was fixed. Each ring stand bore a universal clamp, a small piece of Bessemer 

 rod, and a single burette clamp used to hold a i candle-power incandescent lamp in 

 any desired position. Our experiments were performed with the center of the in- 

 candescent filaments 20 cm. from the bottoms of the experimental boxes, and one 

 fourth of the distance from each end. These boxes were placed in a hood painted 

 dead black, and provided with two curtains, one of which hung from above down- 

 ward and contained a small slit for the observer, the other when in position came up 

 from below to a point 10 cm. above the level of the lamp. This excluded practically 

 all the faint light of the room, and the light was between the animals and the ob- 

 server which made it less easy for them to see him. 



The control box (C) was supplied either with untreated air direct from the pump 

 or with no air. The experimental box (E) and Fig. lA, when experiments with 

 atmospheric humidity were being performed, was supplied with three kinds of air, 

 wet at one end, dry at the other, and untreated or medium air in the center. The 

 dry air was rendered dry by passage through three or more sulphuric acid filters, 

 Fig. iB (see Shackell, '09). Each filter consisted of a Whitehall Tatum museum 

 jar, 45 cm. tall and 9 cm. in diameter at the top. The covers were pieces of plate 

 glass, 13 mm. thick and provided with two holes 25 mm. in diameter. They were 

 clamped into place by ordinary hardware screw-clamps with a 6 cm. opening. The 

 contact points were filed smooth and covered with rubber tubing. The air entered 

 through a glass tube inserted in a rubber stopper in one of the two holes in the cover 

 just mentioned. In order to conduct air to the bottom of the jar, the tube through 

 which it entered was inserted into a second cork fitting inside a larger tube (Fig. iJB) 

 so that when the glass plate was pushed into place, this stopper was pushed inside 

 the glass tube so as to make an air-tight connection. The air left the filters from 

 the top through a short tube inserted in the second rubber stopper. The filter 

 jars were filled with crushed pumice stone, the pieces ranging from 0.5 mm. to 7 mm. 

 in diameter, as indicated by the meshes of the screens used in separating and re- 

 moving the finest and coarsest pieces. This was impregnated with crude sulphuric 

 acid. 



After leaving the drying filters, the air passed through glass wool filters, made by 

 filling ordinary i-liter, wide-mouthed wash bottles \vith glass wool. Tests of the 

 air, made by bubbling the supply from these filters through a methyl-orange solu- 

 tion for a period of 30 minutes, indicated that no sulphuric acid passed into the 

 final delivery pipe. 



