C = D, n /D in = 1.8. The natural grain size distribution curve for the 

 sand is shown in Figure 17. The sand contained both rounded and elon- 

 gated particles. Microscopic examination indicated that most of the 

 particles were porous and had a rough texture. Even the very small par- 

 ticles were porous, and some were even hollow. The specific gravity of 

 solids measured 2.80. The grain size distribution curves for the crushed 

 sand, compared with the natural sand, are also shown in Figure 17. 



The calcareous sand from Florida was the same material used in the 

 previous laboratory study. It was a uniform calcareous sand with a 

 D, n = 0.4 mm and a uniformity coefficient, C = D ftn /D in = 2.8. This 

 sand was finer than the Guam sand and contained about 2% by weight silt- 

 sized particles (finer than sieve No. 200). The grain size distribution 

 curve for the sand is shown in Figure 18. The specific gravity of solids 

 measured 2.72. The sand contained flat pieces of broken shells as well 

 as bulky particles. Under the microscope, the texture of the particles 

 appeared rough. The sand was also tested after crushing. The grain 

 size distribution of the crushed sands, compared with that of the natural 

 sand, are also shown on Figure 18. 



The Ottawa sand is a uniformly graded silica sand with a D n = 0.45 mm 

 and a uniformity coefficient, C = D,_/D. n = 2.0. The grain size distri- 

 bution curve for this sand is shown in Figure 19. The sand consisted of 

 bulky particles that under the microscope appeared to have smooth, 

 polished surfaces. The specific gravity of solids for this soil mea- 

 sured 2.61. The purpose for using the Ottawa sand was to compare the 

 results of the calcareous sand to a typical non-calcareous sand. 



Test Procedures 



The internal friction angle of each sand was measured by means of 

 triaxial compression tests. Test specimens had a cross-sectional area, 

 A , of 10 cm 2 and were prepared in both loose and dense conditions. The 

 loose samples were made by a raining technique where the dry sand is 

 poured through a funnel into a cylindrical sample mold. The distance 

 between the funnel and the accumulating sand was kept constant. Dense 

 samples were prepared by placing the sand in five layers and compacting 

 each layer with a small, hand-held tamper. The isotropic compression 

 behavior was studied when confining stresses were applied to the samples. 

 This was done by observing volume changes under the compression loads. 



All triaxial compression tests were constant-rate-of-strain tests 

 at constant lateral pressure until stress-strain curve passed a peak 

 value. The rate of deformation for all tests was 0.03 inch per minute. 

 The axial load was measured by means of an electronic load cell, and the 

 axial deflection was measured by a strain indicator. 



The majority of triaxial compression tests were performed on dry 

 sand without volume change measurements. However, for each of the cal- 

 careous sands used in this study, a test was made under saturated condi- 

 tions and the sample volume change was measured during drained shear 

 (i.e., no excess pore pressure build-up). This was done to observe the 

 dilatancy characteristics of the sands. 



The friction coefficient of each sand against metal was measured 

 directly by sliding a rigid flat steel plate on the surface of sand. 

 The plate was 3-1/2 inches by 3-1/2 inches, and was placed on a bed of 

 sand 1 inch deep. After the desired normal stress was applied, the plate 

 was subjected to shear loads in small increments until sliding occurred. 



13 



