short "bursts" of vibration (1 to 2 s) to settle the paste back around the vanes. The molds and vanes were then covered with a thin transparent plastic sheet. Four 3-gang molds were filled as rapidly as possible. This usually required about 6 to 7 min.

In vibrating the molds, a vibrator table with frequency of 60 Hz was used with an amplitude of 0.4 mm. The molds were held on the vibrating table by hand-i.e. no clamps were used, hence the amplitude of vibration of the mold was less than that noted on the dial indicator. A minimum of vibration was used in order to avoid as much as possible the bleeding and segregation that may occur with excessive vibration.

The shear tests were normally started at 10 min after making the first specimen. In making the shear test, the load-transfer device was placed on top of a vane and universal joints and spring section lowered onto the load-transfer device. Torque was then applied to the spring by means of a worm gear and the number of degrees of rotation required to cause a shear failure (that is the maximum torque developed) was observed and noted. Measurements were made on the other vanes at various intervals as noted in the figures until the capacity of the apparatus was approached or no more vanes remained to be tested.

Tests were also made of 2 cements using 3 percentage points more water than that required for normal consistency.

A series of tests was also made on 1:2.75 (cement to graded Ottawa sand) mortars using the standard mixing procedure and the amount of water noted in table 1. The mortar was vibrated only a few seconds, cut off, trowelled once, sawed off, and the 1 cm vanes inserted and a "burst" or two of vibration was used to consolidate the mortar around the vanes. One test was made with one of the cements which had a moderate amount of premature stiffening. In this case the sand and cement were mixed dry, the water added in 5 s and the mortar mixed for only 30 s at the medium speed.

A third series of tests on these cements was made using the 2 cm vanes in 4 × 4 × 16-cm molds, placing 3 vanes in each mold.

5. Results of Tests

The results of tests on the neat cement pastes are presented in figures 3, 4 and 5, plotting individual shear-strength values in g/cm2 versus time in minutes on a log-log scale. A series of three straight lines giving the best visual fit were drawn through the values obtained with each of the cements; the values appeared reasonably close to the lines drawn. There were, as may be noted, some values especially in the 10 to 50 min measurements which were erratic.

The slopes of the lines for the different phases of the hardening process were different with the dif ferent cements and were of different duration. Increasing the water content by 3 percentage points delayed by 10 to 20 min the start of the second phase and

about 20 min the start of the third phase with cements 5 and 9.

Indicated on these graphs are the values for Vicat time of set and the times of Gillmore initial and final set values as obtained in table 1. At the time of the Vicat initial set, the shear resistance ranged from 400 to 1200 g/cm2 with four of the cements (Nos 3, 6, 9, and 10); this was shortly after the start of the third phase. With two of the cements, (Nos. 4 and 5) the Vicat time of set came somewhat later. At the time of initial set as determined by means of the Gillmore needle, the six cements had a shear resistance ranging from 1300 to 3200 g/cm2 with an average of about 2100 g/cm2. With the apparatus used to gether with the size of the vanes and depth of the paste, it was not possible to obtain the shear resistance at the time of the Gillmore final set. Extrapolating the third phase lines to the time of final set would indicate a shear resistance of 10,000 to 20,000 g/cm3.

The results of tests of the 1:2.75 mortars are presented in figures 6, 7, and 8. The results obtained for cements 4 and 5 were somewhat erratic, especially in the first and second phases of the hardening process. The time of inflection in the curves obtained with the mortars are close to those obtained in the curves for the neat cement. If the lines drawn through the values obtained on the 1:2.75 mortar are super imposed on the lines drawn through the values obtained with

ess than that of the mortar mixed for the time equired in specifications. Only one value was obained in the third phase and this was close to the ine obtained with the mortar mixed according to pecification requirements.

In figures 13 and 14 the results obtained with th ix cements have been traced on the same graphs. The lines of the first two phases appear to cross to a greater extent than was the case in the third phase. Most of he lines in the third phase were fairly parallel on the og-log plot but were displaced with respect to the ime the phase started.

Although a log-log plot was used to show the overall picture of the hardening process, the first phase may well be presented on a linear plot graph as in figure 15. The early stiffening as measured by means of the vane shear apparatus resulted in straight lines. The slopes of the lines obtained with the different cements differ to some extent and if a shear test is used to measure the plasticity of neat cement paste or

FIGURE 13. Graph indicating the shear strength in g/cm2 versus time in minutes of the six neat cements of normal consistency.

mortar, attention must be given to the time after mixing, or placing, that the test is made.

6. Discussion

Many theories have been proposed relating to the mechanism of the early development of structure in hardening cement pastes. These have been reviewed by Green [3] and more recently by Kondo and Ueda [6]. Among these theories were those of Segalova [19] and Rehbinder [20] who proposed, on the basis of work by a previous Russian research worker, A. A. Baikov, that there were three phases in the hardening process which may overlap in time. The first phase was a colloidization of the cement particles, especially the CA (tricalcium aluminate) after which there is a coagulation of cement particles and reaction. products and finally the development of new hydrated compounds through crystallization which brings about the strengthening.

Sivertsev [21] considered that the initial phase of the setting process was the absorption of water by the

FIGURE 14. Graph indicating the shear strength_in_g/cm2 versus time in minutes of the six 1:2.75 (cement to graded Ottawa sand) mortars of standard consistency.

cement particles, which was followed by formation of micelles which later coagulate into structures. Budnikov and Strelkov [22] have shown that a fibrous material is formed within the first few seconds. These fibrous materials were noted in extracts from cement-water mixtures. The extracts were dried and observed using a microscope.

Schwietes, Ludwig and Jager [23] have presented electron micrographs indicating that ettringite is formed on the CA in the form of needles within a few minutes and lance-like foils of Ca(OH), had also formed on the CaO.

Even after a few minutes contact with water there is (1) an apparent roughening of the cement particles (especially the C,A) [23] making it more difficult for the particles to move past each other in the shear test, (2) a reduction in the amount of free water because of the approximately 31 molecules of water taken up per molecule of ettringite, lowering the water/cement ratio of the paste, and (3) a solution of some of the compounds in cement which may be absorbed on the grains of cement [22]. All or combinations of these factors may be responsible for the slight increase in

FIGURE 15. Graph indicating the shear strengths in g/cm versus time in minutes of neat cement pastes normal consistency.

shear resistance in the first phase of the hardenin process.

The second phase which coincides approximatel with the time of the start of the second increase in hea generated by cements may be, as suggested by Sivertse [21], the formation of micelles which then filled the space occupied by the free water. Increasing th water/cement ratio with cements 5 and 9 increased the time at which the second phase started by 10 t 20 min. The 1:2.75 mortars with a higher water/ce ment ratio had generally a slightly longer first perio than the neat cements with the lower W/C. In som instances, however, the times of the first inflection of the curves were very close. Early investigators in this field have indicated that the solutions are super saturated which may account for the formation of the micelles.

The third phase, starting at 80 to 160 min with dif ferent cements after mixing with water, may indicate the start of chemical bonding of the hydration pro ducts. Cement pats at this stage will normally have lost their sheen, indicating the end of the bleeding period and the continuing uptake of water by the hy dration process. It was of interest to note in compar ing the curves of the 1:2.75 mortars and those of the neat cements, that the time of the start of this phase was, with each of the cements, within 10 to 20 min irrespective of the differences in water/cement ratios and the presence of sand grains in the mortars.

It was observed in conducting the shear test that the method of failure was not always the same. In the first phase the paste or mortar built up in front of the vane and a space was noted in back of each vane. There were no visible cracks extending into the rest of the paste from the edge of the vane. As the paste be came stiffer, as in phase two, cracks were noted from the edges of the vanes in the direction the blades were turning. The V-shaped cracks extended half a centimeter in some instances and the paste or mortar at time of failure was not attached to the back of the