pipe was threaded into the wall of the tank 1 

 inch above the bottom to indicate the level of 

 the free vmter surface within the tank. 



Forty holes 1-1/8 inches in diameter, and 

 arranged in four vertical rows of 10 holes each, 

 were drilled through the walls of each tank to 

 check bulk density of the porous medium. 

 These vertical rows were spaced equally around 

 the tank and the holes within each row were 

 spaced equally from top to bottom. The holes 

 were sealed by expanding plugs made from 

 black rubber stoppers having washers on each 

 face and pierced by bolts. The plugs were com- 

 pressed and expanded when the bolts were 

 tightened. 



The dimensions for the test voids were 

 selected after an examination of field data from 

 six access tubes. These tubes had been installed 

 6 feet deep and showed an extremely large sea- 

 sonal fluctuation in soil moisture. Three of the 

 tubes yielded neutron counts that seemed 

 "reasonable," i.e., percent moisture by volume 

 did not exceed soil pore space when soil was 

 saturated. The remaining three tubes yielded 

 questionable data, i.e., data indicating that the 

 percent moisture by volume exceeded the pore 

 space of the soil. 



Castings were made of the access holes. Be- 

 fore this could be done, the rubber stoppers 

 sealing the bottoms of the access tubes had to 

 be removed. Cement was poured into the tubes 



and 3/8-inch reinforcing rods were used to 

 work the mortar gently out the bottom of the 

 tubes. As the cement flowed out into each of 

 the access holes, the access tube was raised 

 slowly and the reinforcing rod was left in the 

 hole. When the access hole was filled complete- 

 ly with cement, the access tube was removed; 

 and the cement was allowed to harden for 2 

 weeks. Then the castings were excavated (fig. 

 2) and measured. The three access holes yield- 

 ing questionable data had voids, as determined 

 by the castings, that ranged in size from 3 to 6 

 inches in diameter and up to 10 inches long. 



The small and large test voids were made 

 from 10-inch lengths of 3- and 6-inch-diameter 

 aluminum irrigation pipe having wall thick- 

 nesses of 0.010-inch. These test voids repre- 

 sented two symmetrical voids with 1/2- and 

 2-inch annular space around the access tube 

 (fig. 3). 



A heavy pipe frame was erected in the labo- 

 ratory to support an "I" beam which acted as a 

 track for a rolling carriage that carried a 

 hydraulic cylinder. Four lifting bars were at- 

 tached to the top of each tank. When the 

 hydraulic cylinder was attached to the lifting 

 bars a small hydraulic hand pump was used to 

 actuate the cylinder and lift the tank for 

 weighing. 



A coarse sand and a fine sand were used as 

 the porous mediums for the study. Of the 



Figure 2. — Access tube 

 castings were excavated 

 for shipment to laboratory. 

 Some of the six castings 

 showed voids up to 

 6 inches in diameter 

 and 10 inches long. 



3 



