One hundred breeder samples were selected based on their HVI properties (range of length and micronaire). The samples were divided in 4 sub-samples (2 sets for the Old AFIS, 2 sets for the AFIS PRO). For each AFIS instrument, the first set was tested raw with 5 replications of 3,000 fibers, the second set was opened with a non-aggresive fiber blender.

The results show that with blended samples the confidence interval for two slivers is as good as or better than the confidence interval for 5 slivers with the raw material. It means that blending the breeder samples prior to testing could substantially reduce the number of replications needed; therefore, the number of samples tested per unit of time could be significantly increased.

Further evidence of improved performance with systematically blended fibers was provided by the effective utilization of fibers contained in the AFIS slivers fed into the instrument. When raw, unblended fibers were used, it was observed that a maximum of only about 4,000 fibers could be measured from an AFIS sliver. But when the blended slivers were used, 10,000 fibers could consistently be measured from one AFIS sliver. This is because a much larger portion of the raw, entangled fibers were rejected by the AFIS. An important implication of this result is that a single, blended sliver will allow the measurement of many more fibers. This, in turn, will improve both the speed and the reliability of AFIS measurements.