256 Dr Dickson on Diplostemonous Flowers, &c. 



spermum^ the succession may be described, in general terms, 

 as obliquely from side to side. As I have already stated, 

 the two stamens which first appear in Gardiospermum, are 

 those which alternate with sepals 1 and 4, 4 and 2. Now, 

 it may be supposed by some to be a formidable, if not fatal, 

 objection to my views, that two of the supposed accessory 

 stamens should appear before the primary ones to which 

 they belong. At first sight, such a mode of appearance 

 seems very improbable ; yet, when we consider the arbitrary 

 manner in which the disturbing force affects the order of 

 staminal succession, we need scarcely be surprised even at 

 such a result. At any rate, it cannot, d priori, be said to 

 be more improbable that the appearance of primary staminal 

 lobes should be delayed by a disturbing force until after 

 that of their accessory or lateral lobes, than that the appear- 

 ance of a normally older staminal whorl should be delayed 

 until after the appearance of some of the parts of a 

 normally younger one, which must be admitted on the 

 ordinary supposition of there being two genuine staminal 

 whorls. I do not think, therefore, that the case of Cordio- 

 s'permum, although certainly a very strange one, can fairly 

 be urged as invalidating my views. 



While engaged in the attempt to determine the morpho- 

 logical constitution of double staminal whorls, I was led, 

 incidentally, to examine the position of the carpels in some 

 of the Malvaceae. 



I have already stated, regarding the Biittneriacese, that 

 where, in these plants, there are two staminal whorls, 

 the carpels (as alternating with the younger staminal 

 whorl) are superposed to the petals, — e.g. in Lasiopetalum, 

 Bilttneria, 3IelocJiia, &c. ; — but that w^here the androecium is 

 reduced to a single whorl (of simple or compound stamens), 

 the carpels are superposed to the sepals, as in Hermannia 

 and the Dombeyese. 



The researches of Payer leave no doubt that the androe- 

 cium of the MalvacesB consists essentially of a single whorl 

 of five compound stamens, superposed to the petals. With 

 a staminal arrangement so closely analogous to what Baillon 



