Determinations of hardness were made on 324 samples of butter drawn monthly for 3 years from nine factories distributed throughout New Zealand. The results were compared by statistical methods with the iodine values and softening points of the corresponding butterfats which had been determined by Cox & McDowall(4). It was shown that the softness (reciprocal of hardness) was significantly correlated with both iodine value and softening point, which together explained 55 % of the variation in softness observed. Of the remaining variation, 59% was due to seasonal changes other than those in iodine value and softening point and 36 % to differences between factories. It is considered that manufacturing variations were responsible for very little of the seasonal variations, and that the differences between factories were due as much to district differences as to manufacturing differences. It is concluded that only a small part of the variation in softness of the butter samples was due to manufacturing variations, and that at least 80 % of the variation could be attributed to differences in the raw material.