AASHTO T 105:2021 Standard Method of Test for Chemical Analysis of Hydraulic Cement.
Referee Analyses— When conformance to chemical specification requirements is questioned. perform referee analyses as described in Section 4.1.1. The reference test methods that follow in Sections 7 through 22, or other test methods qualified according to Section 5.4, Performance Requirements for Rapid Test Methods, are required for referee analysis. A cement shall not be rejected for failure to conform to chemical requirements unless all determinations of constituents involved and all necessary separations prior to the determination of any one constituent are made entirely by these methods. When reporting the results of referee analyses, specify which test methods were used.
4.1.1. Referee analyses shall be made in duplicate and on different days. If the two results do not agree within the permissible variation given in Table 1, the determination shall be repeated until two or three results agree within the permissible variation. When two or three results do agree within the permissible variation, their average shall be accepted as the correct value. When an average of either two or three results can be calculated, the calculation shall be based on the three results. For the purpose of comparing analyses and calculating the average of acceptable results, the percentages shall be calculated to the nearest 0.01 (or 0.001 in the case of chloroform-soluble organic substances), although some of the average values are reported to 0.1 as indicated in the test methods. When a blank determination (see Note 1) is specified. one shall be made with each individual analysis or with each group of two or more samples analyzed on the same day for a given analyte.
Note I —A blank determination is a procedure that fbllows all steps of analysis, but in the absence of a sample. It is used for detection and compensation of systematic bias.AASHTO T 105 pdf download.