Prom a number of Ginkgo trees kernels were examined. The investigation of the variability of the material was greatly favoured by the large number of stones, in total about 4700. In addition, one tree, grown in the Botanic Garden at Leyden, supplied the material for an investigation of the variability of the female “flowers”, in total about 1700. 1. The number of female “flowers” or rather macrosporangiophores on the brachyblasts (short shoots) proved to be most variable, showing a correlation with the age of the shoot (Table I, fig. 1). 2. A subdivision of the macrosporangiophores into a series of types proved to be possible (Table II, fig. 2-3). 3. A certain relation between the shape of the macrosporangiophore and the shape of the leafy organs from whose axil it arises, was stated. Here again the age of the brachyblasts plays a part. It should be emphasized that the term “abnormality” is misleading. A great number of so-called abnormalities in the macrosporangiophores of Ginkgo prove to form part of a normal series of gradating variations (Table III, fig. 4). 4. There proved to be a relation between the shape of the seed and the shape of the kernels (fig. 5). Oblong seeds give long, pointed stones, while pear-shaped seeds contain club-shaped kernels. Furthermore very small seeds with normally shaped, but very small stones were found. Finally seeds are found in which the pollen-chamber is situated laterally instead of apically. In these seeds the stone is abnormal in shape, its sclerotesta mostly being incompletely lignified. 5. A further point of investigation was the length of the seed stalk (Table IV, fig. 6). This shows a considerable variation, the Leyden material possessing very short seeds stalks, while the Maastricht material had intermediate, that from Slikkerveer long stalks. 6. Finally the variation of the shape of the kernel was investigated. First of all a subdivision into stones with 1, 2, 3 or 4 ribs was made (Table V). The Leyden tree produced relatively many stones of the first group, but four-ribbed kernels are very rare, two- and three-ribbed ones being in the majority. The ratio two-ribbed stones: three-ribbed stones proved to be ± 3 (Table VI). It is probable that the material of Affourtit and La Rivière has been subject to some sort of selection, on account of which their results are not fully trustworthy. 7. There proved to be a strong variation in the angles between the ribs in two- and three-ribbed seed stones (Tables VII and VIII). In the two-ribbed kernels a tendency towards angles of 180° was stated (fig. 7), the most frequent shape being that of the kernel of a prune. 8. The graphic expression of the variability of three-ribbed stones presented some difficulties. To their solution Bakhuis Roozeboom’s triangle-method was chosen (figs 8-11). The most important result is the extreme rarity of regular seedstones with three angles of about 120° (Table IX). 9. It is certainly very remarkable that so ancient a plant as Ginkgo biloba shows such a variability in so many respects.