i2o6 A TEXTBOOK OF THEORETICAL BOTANY 



inferior ovary, two completely distinct whorls of carpels, in which, more- 

 over, the upper contains five to seven carpels with parietal ovules and the 

 lower only three, with axillary ovules. This inversion of the normal order 

 of diminishing numbers in the upward direction is probably due to an 

 invagination of the whole centre of the flower into the axis, for which a 

 parallel may be found in Calycanthus (p. 1138), so that the lowermost 

 group of carpels are morphologically the upper, central group, standing on 

 the sunken apex of the receptacle. 



Spiral flowers, such as Ramincuhis, present a great deal of variation in the 

 number of parts in each zone, as well as in the total number of parts. The 

 number of stamens and carpels is positively correlated with the number of 

 perianth parts, but it may be observed that there is an inverse or negative 

 correlation between the number of stamens separately and of carpels 

 separately, as if there were only room for a limited number of appendages 

 on the receptacle and when more of these are stamens then fewer are 

 carpels, and vice versa. The number of carpels may also be reduced by 

 abortion of the uppermost ones (cf. Caltha, p. 1131). Cyclic flowers show 

 a greater stability of numbers, and in them there is often a close linkage 

 between the number of carpels and the number of stamens, especially the 

 number in the inner whorl of stamens, when there is more than one whorl. 

 Variations in the respective numbers are closely linked. This may be 

 accounted for, if we consider what has been said before about sectorial 

 development in the flower, by the abortion of one or more sectors of the 

 receptacle, involving the disappearance of a corresponding number of 

 stamens and carpels together. Such variations in development are often 

 influenced by nutrition and in starved plants there may be a general reduc- 

 tion of the number of parts, which may however affect the stamens more 

 than the less numerous carpels. 



When there are only two carpels, these, in zygomorphic flowers, always 

 lie in the plane of symmetry of the flower, whether this is, as usual, the 

 vertical plane or an oblique plane, as in Solanaceae. 



The cohesion of carpels occurs, as we have seen, in various degrees of 

 completeness. In some flowers a group of carpels which are essentially 

 separate from each other, may be so closely pressed together, as in Nigella 

 and Bntormis, that it requires close examination to ascertain that no organic 

 fusion exists. The carpels in Aqiiilegia and in the Saxifragaceae are united 

 only at the base. In Apocynaceae they are united only at the top. In 

 Centrolepis they are united one above the other in an alternating, vertical 

 series; while in Pandamis they unite only at the fruiting stage. In the great 

 majority there is organic union between the lateral walls, but each carpel 

 retains a distinct, closed loculus, the fused lateral walls forming the dividing 

 septa between the loculi (Fig. 1177). Such compound ovaries are pluri- 

 locular. An anomalous condition exists in Leptodermh, one of the Rubia- 

 ceae, in which the septa are present but are trellised instead of being solid, 

 so that the ovary is technically plurilocular but is actually unilocular. A 

 further degree of cohesion is that in which the lateral walls of each carpel 



