72 ON SEEDLINGS 



seed, as in the Nut (fig. 117, o), Beech (Fagus), &c., or, as in 

 the Oak (fig. 116, o), near the base. Their presence appears 

 to indicate that these species are descended from ancestors 

 the fruit of which was composed of several cells, each with 

 more than one seed a state of things therefore very unlike 

 the present, and when the anatropous condition would be an 

 advantage. If this view be correct, the structure of the fruit 

 in the Nut, Beech, &c. becomes peculiarly interesting, because 

 it represents a case in which the present arrangements are 

 not those in all respects most convenient to the plant, and 

 renders it probable that the same explanation may apply to 

 other cases of difficulty. 



We meet indeed with a great many cases in which a larger 

 or smaller number of the ovules, often all but one, fail to 

 become developed. In some of them we may perhaps see a 

 provision to increase the chances of fertilisation. There are 

 others, however, in which this explanation will not hold good. 

 In Ptelea, for instance, the ovary contains 2-3 cells, each 

 containing two ovules, one inserted rather above the other, 

 and the inferior ovule constantly fails to develop. In this and 

 other similar cases we can hardly doubt that the existence of 

 this second, and now functionless, ovule carries us back to a 

 time when the ancestors of our present Ptelea habitually pro- 

 duced two seeds. 



In Paliurus the fruit normally contains three, but some- 

 times only two, loculi, each with a single seed. Again, in 

 Myagrum the ovary is spuriously 3-celled, but the two outer 

 cells produce no seed. In Haemanthus there are three cells, 

 each containing a seed, only one of which, however, comes to 

 maturity. In Convallaria there are three cells, each with two 

 ovules, but only one of the six is generally developed. In 

 Phillyrea and Canarium there are two ovules, only one of 

 which is developed ; and we meet with similar cases in 

 Gyrinopsis, Jasminum, .ZEsculus, Cordia, and many others. 



From the point of view as to the origin of these differences, 

 the variations in seedlings offer great interest. For example, 

 out of 135 seedlings of Lepidium sativum (which, as already 

 mentioned, differs from the rest of the genus in having tri- 

 partite cotyledons), no less than 25, or as much as 18^ per 



