1 - The present paper deals with the results of studies on pollen fertility, number of seeds per fruit and frequency of polyembryony in Citrus. 2 - The pollen may be divided into four groups of which the first consists of the living and functional grains, colouring deeply with carmín acetic, germinating in sugar solutions fairly well. The remainny three types : half filled grains, empty grains, and small deformed grains are all inviable. 3 - The grains are generally spherical, 15 to 40ix in diameter. The smallest grains were found in Citrus hystrix and Citrus mitis (fig. 3 and 4), while citron and the grape-fruits have the largest grains (fig. 1, 2 and 5). 4 - Size and form are reasonably constant, except the hybrid "Citran-gequat", "limão galego" without spines, orange "Ruby" (fig. 6 to 8). 5 - Percentages of viable pollen grains are given in quadros IV to X. Tahiti Lime, probably a triploid, is completely sterile. "Baía" orange (Washington Navel orange) is completely male sterile, while the fruits of it may have some seeds. The two seedless varieties of "lima" and "pera" have 37,1 and 9,3% of viable pollen respectively. Regarding pollen viability oranges are the most variable. They show from 0 to 90% of good pollen (quadros IV and V). In lemons and limes the percentages vary from 50% and 80% (quadros VI to VIII). Tangerines, with exception of Satsuma owari, have generally more than 60% of good pollen. The grape-fruits, except the case of "Marsh-seedless", are characterized by a relative high pollen fertility. 6 - There is a certain variation of pollen fertility due to the external conditions. It seems to exist also a certain effect of the rootstock, on viability of pollen grains, as shown for the rootstock of "sour orange" and "rough lemon". 7 - The number of seeds per fruit was also determined and is given in the respective quadros (IV-X). 8 - There does not exist any consistent difference between open pollinated and selfed flowers in regard to the number of seeds per fruit (quadro XII). In 26 cross pollinations (quadro XIII), 6 caused an increase, 11 a decrease in the number of seeds. Especially interesting in this respect was "Baía" orange (Washington Navel), where, for instance, the cross pollination with the orange "Barão" gave in average 7 seeds per fruit, while in the cross with "Hamlim" the seed number was only 2,75 per fruit, this difference being highly significant. 9 - It is a well known fact that nucellar embryony is quite common in Citrus. The frequency of polyembryony has been determined for a large number of types. Even if an embryo is formed, we may not be certain whether it is produced by an adventicious growth of the nucellus or from a normally fertilized egg, as shown for instance for Citrus mitis x Poncirus trifoliata. (Fig. 15 and 16). Both twin embryos may be apogamous or one only, the other being in this case an hybrid. 10 - There exists a slight positive correlation between number of seeds and percentage of good pollen, with r = -f- 0,53. It should not be forgott...