IMPROVING QUALITY-CONTROL PROCEDURES FOR SEED-BORNE PATHOGENS BY TESTING SUB-SAMPLES OF SEEDS

Quality control procedures for seed-borne pathogens may involve testin g a single sample of seeds for the presence of the pathogen, but the p robability of rejecting seed-lots with acceptable infection levels can be high. If multiple sub-samples are tested and the seed-lot is rejec ted when any of the sub-samples gives a positive result then the proba bility of inadvertantly rejecting acceptable seed-lots is unchanged. H owever, this probability can be reduced, without reducing the probabil ity of rejecting unacceptable seed-lots, by requiring more than one of the sub-samples to give a positive result before rejecting the seed-l ot. We present the underlying statistical theory behind this result an d discuss the implications for seed testing procedures.