Characterization of the entrained air void system of selected cores was performed using the modified point count procedure of ASTM C 457 . Data were obtained from a slab sawn from the center of each core, carefully smoothed and polished to present a surface suitable for microscopical examination. The prepared surface was then stained using a potassium permanganate, barium chloride stain per Poole and Thomas  to highlight air voids filled with ettringite (Figure 8). The point count was performed using a semi-automated linear traverse system  with an additional point count tabulator. Each data set represents a total traverse length of at least 2540 mm, in accordance with ASTM C 457 requirements. Uncertainties in the measurements are estimates of two standard deviations, corresponding to a 95 % level of confidence. The linear traverse analysis counted all voids, that is, all effective and those rendered ineffective due to filling.
Figure 8. Potassium permanganate and barium chloride staining colors sulfate phases purple facilitating their identification. This allows discrimination of both the original and effective entrained air void systems.
The second tabulator was used to track the filled air voids, permitting recalculation of the data to reflect the effective entrained air void system. This facilitated counting of two air void systems, that of the concrete as placed, and that of the concrete at the time of sampling. These values differ because of the subsequent filling of some of the entrained air voids with ettringite. Criterion for distinction between an ineffective and effective void was that the void must appear completely filled. Voids rimmed or partially filled with secondary mineralization, and those with open cores were considered effective as entrained air voids.