478 Saunders.—A Reversionary Character 
six component valves, whereas in T. painstre the three valve carpels separate 
from below upwards, remaining attached by their apex to the tri-lamellate 
frame formed by the three reduced carpels. 1 
Having thus shown by reference to some half-dozen genera that we 
meet with precisely the same dimorphism and division of labour between the 
carpels in the Orchidaceae as in the Rhoeadales, and that the conception of 
the ‘ commissural 5 stigma and the ‘ false ’ partition is not more a reality in 
the one case than in the other, we must leave a more complete survey of 
this family for future investigation. It may be noted in passing, how¬ 
ever, that the above interpretation is applicable equally to the neighbouring 
small family of the Burmanniaceae. 
It remains to review briefly the principal conclusions to be drawn from 
the facts here investigated. 
Summary of Conclusions. 
General. 
1. Evolution in the Rhoeadales has been accompanied by reduction and 
consolidation of the members of the gynoecium, leading to the production of 
two kinds of carpels, the ‘ hollow ? or valve type and the ‘ solid ’ type. 
2. When the two kinds of carpel coexist, they are disposed regularly 
and alternately, except in the rare cases where a member of the gynoecium 
is undeveloped. 
3. The solid carpel is recognizable externally by its double contour line, 
as is well seen, e. g., in the typical Crucifer. Contiguous valve carpels on the 
other hand give rise to a single-line suture. 
4. The morphological transformation from the valve to the solid type 
has been accompanied by redistribution of the carpellary functions. 
5. When both types of carpel are present, the ovuliferous function is 
almost always taken over by the more recently evolved solid carpel, and the 
more primitive valve carpel becomes sterile. 
6. The receptive (stigmatic) function is not rigidly associated with one 
particular type of carpel, but may be carried out by either the one or the 
other or by both. All three dispositions may be found within the same 
family, as is the case in the Cruciferae. 
7. An alternate arrangement of solid and hollow carpels may act 
advantageously by facilitating a more complete dehiscence of the fruit than 
usually occurs where valve carpels only are present. In the latter case 
separation generally begins from above, and is only partial. 
1 The suggestion was made by Gaertner as long ago as 1791, that the three lamellae alter¬ 
nating with the three full-sized carpels might perhaps represent three other members which had 
been suffocated by their neighbours (De Fruct., vol. ii, p. 26). As Gaertner mentions that the 
stigmas are sometimes three and sometimes six, possibly we have here a parallel in this respect to the 
variability in stigma formation which occurs in Eschscholzia californica (see p. 471). 
