297. As in the case of sand, the shape of the fragments and the degree of uniformity in size have an important effect on the proportion of voids in the mass, and all of these elements affect the value of broken stone for use in concrete. As in mortar each grain of sand should be completely covered with cement, so in concrete should each piece of stone be completely covered with mortar. As the pieces in a given volume of broken stone will have a smaller total superficial area when the fragments are large than when they are small, we should conclude that the larger fragments will require less mortar or be more thoroughly coated with a limited amount. From the same point of view we should expect that round fragments would require less mortar than those of irregular shape.

It is found however, in practice, that these theoretical considerations must be modified to correspond with the facts.

Table 64. Voids In Broken Stone And Gravel Varying In Granulometric Composition

Note. — Stone jarred down in measure for all trials. K passed holes 1/4 inch square, failed to pass holes 1/10 inch square. V " 1/2 " " 1/4 "

F " 1 " " 1/2 "

M " 2 " " 1 "

C " 3 " " 2 "

When the pieces of broken stone are too large, they do not bed themselves well in the matrix of mortar, but become wedged one against another, leaving voids in the concrete. While round fragments have a small superficial area in relation to their volume, have a small percentage of voids, and pack together readily, yet they are lacking in ability to form a good bond, and hence do not give the best results.

298. Relation Of Size Of Stone To Volume Of Voids

As illustrating the effect of the size of fragments and granulometric composition of stone on the volume of voids, Table 64 gives a number of results obtained at St. Marys Falls Canal.

Table 65 gives some of the results obtained by M. Feret in similar tests.1

Table 65. Size Of Stone And Volume Of Voids

The percentage of voids in a mass of broken stone of uniform size should be independent of what the size may be, and the first few lines in Table 65 show this to be nearly the case with the four samples tested. It is seen from both tables that the more complex mixtures give smaller percentages of voids, and that for all sizes the voids are much less in the gravel than in the broken stone.

1 " Sand and Stone Used for Cement Mortar and Concrete," by M. Feret. Abstracted in Engineering News, March 26, 1892.

Table 66. Strength Of Concrete. Varying Size Stone

size "G" of broken stone passed a ring 60 mm. (2.4 inches) in diameter and was held by a ring 40 mm. (1.6 inches) in diameter; "M" passed 40 mm. (1.6 inches) ring and was held by a ring 20 mm. (0.8 inch) in diameter; while "F" passed the 20 mm. (0.8 inch) ring and would not pass a ring 10 mm. (0.4 inch) in diameter.

The following conclusions are drawn from this table: (1) In each block the lower layers, which had been submitted to longer continued ramming than the upper layers, offered a greater resistance. (2) The mean resistance varied according to the granulometric composition of stone used, and was greater with the increasing proportion of large stone in each block. Since the amount of mortar used was in all cases equal to the volume of voids in the stone, the effect of voids on the strength was not noticeable.

299. M. Feret's Experiments

To show the effect of the variation in sizes of fragments on the strength of the concrete made, M. Feret experimented with four mixtures of three sizes. The proportions used in the mortar were one part by weight of Portland cement to three parts of Boulogne gravel, gaged with an amount of water equal to seventeen per cent, of the total weight of cement and sand. The volume of mortar used in each case was made equal to the volume of the voids in the stone. The concrete was thoroughly mixed and then rammed into a large cylindrical mold. After four months' exposure to the air, twelve cubes were cut from the cylindrical block, four cubes being cut from each of three consecutive horizontal layers. These cubes were placed in sea water and crushed after one month, being then five months old. The results of the tests are given in Table 66.

300. Further Experiments

Tables 153 and 155 give the results of some experiments made under the author's direction to test the effect of size and character of broken stone. In these tests the proportions are generally 35 pounds of cement to 105 pounds of sand and 3.75 cubic feet of broken stone, the stone being measured after jarring it down in the vessel. The amount of mortar made was sufficient to fill the voids in the stone when the latter did not exceed about thirty-three per cent (§§452, 454).

It is seen that, in general, a higher result was given by mixtures of various sizes than by any one size alone, and the fine stone gave higher results than the coarse. In these tests the effect of voids is shown, since in some cases there was not sufficient mortar to fill the voids.

301. Gravel Vs. Broken Stone As Aggregate

The elements entering into the analysis of the superiority of one kind of aggregate over another are given above, but since the question of the relative merits of gravel and broken stone is so frequently discussed, a word may be added here to show the special points involved in such a comparison.

Gravel is composed of hard, rounded pebbles, the surfaces of which are usually quite smooth. On account of the manner of its formation and occurrence, the sizes of the pebbles are usually graded from coarse to fine. Occasional beds of gravel are found, however, in which the sizes of the several fragments are nearly the same. In broken stone the fragments are angular and usually have rough surfaces, though the degree of roughness depends upon the kind of stone. The sizes of the fragments as they come from the stone crusher vary from coarse to fine, but by regulating the crusher jaws and by screening, any desired size may be obtained.