tability of the variables affecting the numerical productivity of the sows. The herd of origin accounted for about 5 per cent of the variance of litter size and for 30 to 5 j per cent of that of the lactation length. Estimations of the repeatability of the litter size varied from o.ig to o. y those of the other variables were less homogenous. Within-litter linear correlations between variables showed that the litter size was independent of the weaning-fertilization interval, but tended to

more »

... se with lactation length. The curves representative of the polynomial regression equations accounting for the variations of the litter size at weaning in terms of litter size at birth show that the latter variable increases up to a maximum of 17 piglets born for Large White and French Landrace, 15 for I'ietrain and16for Belgian Landrace. However, a piglet has a maximum of chances for surviving until weaning