We checked whether average frequencies calculated were different. The variable analyzed was the proportion (p) (p = X/n), where X was the number of abnormal plants and n the number of plants observed by culture age. A 3r confidence limit for binomial distribution was calculated using the formula