SAMPLING IN BLENDING VALIDATION

It is shown by a short statistical experiment, that sampling size in blending validation prior to the point where blending is as complete as it can be, is related to the size of the sample by being inverse function of the weight of the sample. The use of small thieves of the size of the dosage unit are apt to cause mechanical separation, and bias the results from the validation experiments. If the powder mixture is adequatly blended, then scaling can be carried out by considering the mass to be of a binomial distribution. The article is a preliminary article and will be followed by work done on larger matrices. It is noted that it is educational in nature, attempting to show in an easy manner what the parameters used and what the effects are. In the work of blending, experimental results usually have large variation, and for such considerations, simulation, as presented here, is a more fruitful approach.